CEA-TCB (cibisatamab), HLA-A2 WT1-TCB (RG6007), CD19-TCB and DP47-TCB were produced internally in the 2+1 TCB format previously described (Bacac et al, CCR for CEA and CD20-TCBs). Dasatinib (S1021) was purchased from Selleckchem.

The SKM-1 cell line (DSMZ# ACC547) is a human acute myeloid leukemia cell line. The cells were cultured in RPMI (11875101; Gibco) containing 20% FBS (26140079; Gibco) and split every 3 to 4 days (to 0.6 million cells/mL). To ensure sufficient MHC I levels, cells were used in an assay 1–2 days after passaging. For the Incucyte experiment, SKM-1 labeled with NucLightRed (NLR) were used.

The NLR-labeled A375 cell line is an adherent melanoma cell line which is HLA-A2 and WT1 positive. It was transduced with a vector coding for histone-staining red fluorescent protein. The cells were cultured in DMEMF12 (11320033; Gibco) containing 10% FBS (26140079; Gibco) supplemented with 6 µg/mL puromycin and split every 3 to 4 days (to 20,000 cells/cm²). Cells were plated 1 day prior to the assay and pulsed with RMF peptides 2 hours before starting the assay.

The NLR-labeled MKN45 cell line is an adherent human gastric cancer cell line, which expresses high levels of the CEA antigen. It was transduced with a vector coding for histone-staining red fluorescent protein. The cells were cultured in RPMI Glutamax (61870036; Gibco) containing 10% FCS and split every 3 to 4 days (50,000 cells/cm²). The cells were plated 1 day prior to the assay.

The SU-DHL-8 cell line is a human large cell lymphoma cell line derived from peritoneal effusion in a 59-year-old male Caucasian patient (ATCC catalog number CRL-2961). The cells were cultured in RPMI (11875101; Gibco) containing 10% FBS (26140079; Gibco) and split every 3 to 4 days (to 0.8 million cells/mL).

Cell line authentication was performed at Microsynth.

Peripheral blood mononuclear cells (PBMCs) were isolated from buffy coats donated by healthy donors (blood donation center in Zürich, in accordance with the Declaration of Helsinki) by Ficoll density gradient. Briefly, blood from buffy coat was diluted 1:1 with PBS and about 25 mL was layered on 15 mL of Ficoll (17-5442; GE-Healthcare) and centrifuged for 30 min at 2000 rpm without break. Lymphocytes were collected with a 10 mL pipette in a 50 mL tube, rinsed with PBS, and successively centrifuged at 1700 rpm (5 min), 1500 rpm (5 min), 1100 rpm (10 min) and 900 rpm (10 min) to remove remaining platelets.

PBMCs were counted and then adjusted to either 0.4×10⁶ cells/mL, 1.0×10⁶ cells/mL, 2.0×10⁶ cells/mL or 6.0×10⁶/mL in assay medium. Then 50 µL of the cell suspension was transferred to the wells of the assay plates, corresponding to 20,000, 50,000, 100,000 or 300,000 cells/well.

HLA-A2 WT1-TCB, CEA-TCB, DP47-TCB and CD19-TCB were diluted in assay medium. A series of eight dilutions (1:10) was prepared by transferring and mixing 100 µL of 400 nM TCB solution to the subsequent wells containing 900 µL of assay medium. A 20× dasatinib solution was prepared in PBS from a 10 mM DMSO stock solution and transferred into the wells (10 µL/well).

For in vivo administration, dasatinib was formulated in 10% DMSO, 30% PEG300, 5% Tween 80% and 55% H₂O in a stock solution of 10 mg/mL

One day before the assay, adherent NLR-labeled A375 or NLR-labeled MKN45 target tumor cells were detached from the plate using 0.05% trypsin (25300096; Gibco). Cells were washed with PBS and the counts of viable cells (>90%) were determined by Trypan Blue staining using an EVE cell counter. Cells were re-suspended in pre-warmed assay medium (37°C) to obtain a cell density of 50,000 cells/mL. Then 100 µL of the cell suspension was transferred into a 96-flat-bottom well plate, corresponding to 5000 target cells per well.

On the day of the assay, SKM-1 or SU-DHL-8 tumor cells were washed with PBS and the counts of viable cells were determined by Trypan Blue staining using an EVE cell counter (>90%). If required, the cells were labeled with Cell Trace CTV (C34557; Thermo Fisher) or CFSE (C34554; Thermo Fisher). Cells were re-suspended in pre-warmed assay medium (37°C) to obtain a cell density of 200,000 cells/mL. Then 100 µL of the cell suspension was transferred into a 96 U-bottom well plate, corresponding to 20,000 target cells per well.

For the Incucyte experiments, NLR-labeled SKM-1 cells were attached to a 96-well plate using rectronectin from Takara. Flat-bottom 96-well plates were coated with 50 µL of a rectronectin stock solution of 10 µg/mL (45 min, RT). The plates were washed with PBS, and the cells were plated (45 min, 37°C) before the assay.

SKM-1 tumor cells were washed once with sterile PBS and labeled with Cell Trace CTV (C34557; Thermo Fisher) or CFSE dye (C34554; Thermo Fisher) (5 µM, 20 min at RT), washed with RPMI+20% FBS and counted.

In assays with 20,000 SKM-1 tumor cells/well, 50 µL of the effector cell suspension (20,000, 100,000 or 200,000 cells/well) were added, followed by 50 µL of the antibody solutions. The final E:T ratio was 1:1, 5:1 or 10:1 and the total volume per well was 200 µL. The assay plates were covered with lids, and placed in the incubator, 37°C, 5% CO₂.

The assay medium of tumor cells was replaced with fresh medium (100 µL/well). Then 50 µL of the effector cell suspension (50,000 cells/well) followed by 50 µL of the antibody dilutions were added to the assay plates containing 5000 adherent target cells/well. The E:T was approximately 10:1 and the total volume per well was 200 µL. The assay plates were incubated in the Incucyte for measurements of total red area/well every 3 hours, at 37°C, 5% CO₂.

At the assay endpoint, PBMCs were washed twice in PBS (1500 rpm, 5 min, RT) and stained with the following markers: CD4 (APC-Cy7, 317418; Biolegend), CD8 (BV605, 344742; Biolegend), CD25 (BUV395, 564034; BD), CD69 (PE, 310306; Biolegend) and Live Dead Near Infra Red (NIR) (L10119; Thermo Fisher) for 30 min at 4°C in FACS buffer. For blood staining, 25 µL of blood was lysed twice using BD Pharm Lyse buffer (555899; BD) (200 µL, 10 min, RT) and stained with the following markers: CD45 (Alexa 700, 304119; Biolegend), CD20 (APC, 302309; Biolegend) and Live Dead NIR (L10119; Thermo Fisher) for 30 min at 4°C in FACS buffer. Cells were then washed twice in FACS buffer (1500 rpm, 5 min, RT) and re-suspended in 100 µL/well FACS buffer for analysis. Sample acquisition was performed using an HTS plate reader connected to a BD Fortessa Flow cytometer.

To block cytokine secretion, 20 µL of culture medium containing 1/150 (final concentration: 1/1500) Brefeldin A (Golgiplug, 555029; BD) and 1/100 (final concentration: 1/1000) Monensin A (Golgistop, 554724; BD) and CD107a (Alexa 647, 562622; BD) was transferred to each well for 24 hours. At the assay endpoint, PBMCs were collected, washed twice in PBS (5 min, 1500 rpm, RT) and stained with a mix of antibodies to the following surface markers: CD4 (BUV395, 564724; BD), CD8 (BV605, 344742; Biolegend) and live dead NIR (L10119; Thermo Fisher) in PBS (30 min, 4°C, no light) in PBS. PBMCs were centrifuged (5 min, 1500 rpm, RT) and fixed with Cytofix/Cytoperm buffer (554722; BD) (80 µL/well, 30 min, 4°C). PBMCs were centrifuged and then washed with Perm/Wash buffer (554723; BD) (5 min, 1500 rpm, RT). PBMCs were incubated in Perm/Wash buffer (30 min, 4°C, no light) and then stained with a mix of antibodies to cytokines: TNF-α (APC-Cy7, 502944; Biolegend) and IFN-γ (BV795, 612845; BD) in Perm/Wash buffer (50 µL/well, 30 min, 4°C, in the dark). PBMCs were centrifuged and washed with FACS buffer (5 min, 1500 rpm, RT). Last, they were re-suspended in 100 µL FACS buffer and sample acquisition was performed using and HTS plate reader connected to a BD Fortessa Flow cytometer.

Cytokines were analyzed in the culture supernatants from the killing assays (stored at −80°C) by Luminex using a human8Plex Assay kit (M50000007A; Bio-Rad) and additional reagents for IL-1β (171B5001M; Bio-Rad) and MCP-1 (171B5021M) measurement. Pre-diluted supernatants were incubated with beads in a 96-well filter plate (1 hour, 800 rpm, RT, in the dark). The plate was washed twice using a vacuum manifold and the detection antibody solution was added (1 hour, 800 rpm, RT, no light). The plate was vacuumed and washed twice and the streptavidin solution was added (30 min, 800 rpm, RT, in the dark). The plate was vacuumed and washed twice and the samples were re-suspended in assay buffer. Sample acquisition was conducted using the Luminex equipment from Bio-Rad.

Humanized NSG mice were ordered from the Jackson Laboratory. The Cantonal Veterinary Office, Zurich, Switzerland, approved the protocol (ZH225-17) in accordance with the Swiss Animal Protection Law. One day before treatment, humanized NSG mice were randomized based on their T cell counts into three groups of four mice. One group was treated with dasatinib (50 mg/kg, orally) 1 hour before injection of CD19-TCB (0.5 mg/kg, intravenously) on day 0 and again 5 hours and 8 hours after injection. On days 1 and 2, dasatinib was given twice per day with intervals of 10–11 hours. Blood was collected by tail-vein bleedings or by terminal retro-orbital bleeding at 72 hours.

Flow cytometry data were analyzed using FlowJo V.10. Cytokine data were analyzed using the Bio-Plex software. GraphPad Prism V.8 was used to generate the graphs and for statistical analysis. For dose–titration curves, AUC were calculated and used for statistical comparison. Data are shown as means with SD or SEM or as individual curves. The statistical tests used are indicated in the figure legends for each experiment.

To assess the inhibitory effect of dasatinib on TCB-mediated target-cell killing, PBMCs were co-cultured with NLR-labeled SKM-1 cells and HLA-A2 WT1-TCB, in medium supplemented with escalating concentrations of dasatinib. The Incucyte system was used to capture the loss of red fluorescent protein signal over time as a readout of target-cell killing. A concentration of 100 nM (48.8 ng/mL) and 50 nM (24.4 ng/mL) dasatinib resulted in 92.2% and 95.5% inhibition of target-cell killing induced by 10 nM HLA-A2 WT1-TCB (figure 1A and table 1), while not directly affecting NLR-labeled SKM-1 growth (online supplemental figure 2A). A concentration of 25 nM (12.2 ng/mL) and 12.5 nM (6.1 ng/mL) dasatinib resulted in 87.1% and 80.2% inhibition of target-cell killing for 10 nM HLA-A2 WT1-TCB (table 1). The lower concentration of 6.25 nM dasatinib combined with 10 nM HLA-A2 WT1-TCB only partially inhibited killing (figure 1A and table 1). Similarly, 100 nM and 50 nM dasatinib significantly prevented HLA-A2 WT1-TCB-induced SKM-1 killing as well as T cell proliferation and activation in a killing assay using CSFE-labeled SKM-1 tumor cells co-cultured with PBMCs and HLA-A2 WT1-TCB (figure 1B–E and online supplemental figure 4A,B). Dasatinib did not affect T cell nor SKM-1 cell viability, nor the target expression on SKM-1 cells (online supplemental figures 2A,D,E and 3). Moreover, treatment with a concentration of dasatinib above 25 nM totally prevented the release of IFN-γ, IL-2 and to a lower extent TNF-α (figure 1F–H and online supplemental figure 8A–D). The lower concentration of 12.5 nM and 6.25 nM dasatinib decreased but did not fully suppress cytokine release (figure 1F–H). Overall, these data show that dasatinib can fully prevent T cell mediated target-cell lysis triggered by PBMCs stimulated with HLA-A2 WT1-TCB at in vitro concentrations of 50 nM and above. The inhibitory effect of dasatinib on HLA-A2 WT1-TCB-induced T cell cytotoxicity and cytokine release was confirmed using another target cell line (A375) for HLA-A2 WT1-TCB. (online supplemental figures 1 and 2B).

To evaluate if dasatinib could act as a rapid and potent inhibitor of activated T cells, we first stimulated PBMCs cultured with SKM-1 tumor cells and HLA-A2 WT1-TCB for 24 hours before adding 100 nM dasatinib to the co-culture (figure 2A).

The expression of CD69 and CD25 on CD8⁺ and CD4⁺ T cells at 24 hours showed a partially activated phenotype for T cells stimulated with HLA-A2 WT1-TCB in the absence of dasatinib (figure 2B,C). Along with activation of T cells, IFN-γ, TNF-α and IL-2 were also found in the culture supernatants after 24 hours of stimulation (figure 2D–F).

Following a further 24 hours of incubation, this time in the presence of 100 nM dasatinib, the expression of the early activation marker CD69 and the late activation marker CD25 on CD4⁺ and CD8⁺ T cells at 48 hours were lower than those measured at 48 hours in absence of dasatinib (figure 2B,C). Thus, treatment with 100 nM dasatinib rapidly inhibited further induction of activation markers in pre-activated T cells.

We also measured the cytokine levels in the killing assay supernatants at 48 hours, to assess the impact of dasatinib on T cell-mediated cytokine release. No differences were observed for IFN-γ, TNF-α and IL-2 levels measured at 24 hours and 48 hours, as opposed to the dasatinib-untreated control where cytokine levels had largely increased at 48 hours (figure 2D–F). This indicated that the addition of 100 nM dasatinib at 24 hours had rapidly prevented the release of cytokines by activated T cells.

We also assessed T cell proliferation 120 hours after addition of 100 nM dasatinib in the killing assay, measuring the CTV dye dilution peaks by flow cytometry. The treatment with 100 nM dasatinib decreased the proliferation of CD4⁺ and CD8⁺ T cells induced by 10 nM HLA-A2 WT1-TCB, with a stronger effect on CD4⁺ T cells (online supplemental figure 4C). In addition, CD4⁺ and CD8⁺ T cell counts were significantly higher in untreated cultures than in those treated with 100 nM dasatinib (figure 2G). Dasatinib appeared to impact more strongly the CD4⁺ than the CD8⁺ T cell proliferation (figure 2G and online supplemental figure 4C). These results showed that 100 nM dasatinib added to the killing assay inhibited TCB-induced T cell proliferation with an apparent stronger impact on CD4⁺ than on CD8⁺ T cells.

As demonstrated in these experiments, dasatinib treatment rapidly resulted in the blockade of T cell activation, cytokine release and proliferation indicating a loss of T cell functionality.

To assess whether dasatinib could efficiently prevent TCB-mediated target cell killing by activated T cells, we set up an in vitro killing assay with two stimulation steps, in an attempt to mimic an ON/OFF switch of T cell cytotoxicity. PBMCs were first activated with HLA-A2 WT1-TCB for 20 hours in the presence of SKM-1 target cells labeled with CFSE in the absence of 100 nM dasatinib (ON). The PBMCs were then washed and re-stimulated with the TCB on SKM-1 cells labeled with CTV in the presence of 100 nM dasatinib (OFF switch). The use of CFSE-labeled and CTV-labeled SKM-1 tumors allowed to differentiate the tumor cells used in the first or second stimulation by flow cytometry (figure 3A).

The first treatment with HLA-A2-WT1-TCB induced an upregulation of the early and late T cell activation markers CD69 and CD25 on CD8⁺ and CD4⁺ T cells (online supplemental figure 5), as well as the killing of CFSE-labeled SKM-1 target cells (figure 3B,C), while a non-targeted TCB, DP47-TCB, did not show any activity (online supplemental figure 6). T cells were therefore activated by HLA-A2 WT1-TCB and functional before the addition of dasatinib in the system. Following the second stimulation in the presence of 100 nM dasatinib, 87.30% of CTV-labeled SKM-1 cells were alive while only 2.04% were still alive following re-stimulation in the absence of dasatinib (figure 3B). Thus, addition of dasatinib on the second stimulation prevented HLA-A2 WT1-TCB-mediated killing (figure 3B,C). In addition, the release of IFN-γ, IL-2 and GM-CSF (known to be produced by T cells) and of TNF-α, IL-6 and IL-8 (produced by T cells and monocytes) was fully inhibited on re-stimulation in the presence of 100 nM dasatinib (figure 3D,E).18 19 This result emphasizes that dasatinib can switch off activated T cells, rapidly blocking TCB-mediated cytokine release as well as T cell cytotoxicity.

To investigate how dasatinib could prevent T cell cytotoxicity, we measured the expression of CD107a by intracellular staining as a readout for T cell degranulation (online supplemental figure 7A) after stimulation with 10 nM HLA-A2 WT1-TCB in the presence or absence of 100 nM dasatinib. The addition of dasatinib prevented T cell degranulation, as indicated by the inhibition of CD107a (online supplemental figure 7B,C). This result shows that dasatinib can prevent T cell degranulation and potentially the release of perforin and granzyme B that mediate the killing of target cells.

We then verified whether the effect of dasatinib was reversible on its removal. To this aim, we set up a killing assay with repeated stimulations in the presence or absence of dasatinib and followed the killing kinetics using the Incucyte system (figure 4A). After each stimulation, effector cells were washed and re-stimulated on fresh NLR-labeled MKN45 target cells with 1 nM CEA-TCB in the presence or absence of dasatinib, in order to mimic an OFF/ON/OFF or ON/OFF/ON switch. A dose titration of dasatinib was conducted in this system, showing that 100 nM and 50 nM significantly prevented the killing of NLR-labeled MKN45 cells by 1 nM CEA-TCB (figure 4B and online supplemental figure 2C). In addition, dasatinib blocked CEA-TCB-mediated cytokine release (online supplemental figure 9).

Adding 100 nM dasatinib in the co-culture during the first stimulation resulted in the inhibition of target cell killing, which was then reversed after dasatinib removal for the second stimulation (OFF/ON) (figure 4C). IFN-γ, IL-2 and TNF-α were not detected in the supernatants after the first stimulation in the presence of 100 nM dasatinib, indicating a full inhibition of T cell derived cytokine release (figure 4E). Removal of dasatinib after 3 days and re-stimulation with 1 nM CEA-TCB resulted in the release of IFN-γ, IL-2 and TNF-α, indicating that T cell functionality was restored on dasatinib removal (figure 4E). Subsequent addition of dasatinib at day 6 again inhibited T cell cytotoxicity until day 9 (figure 4C), showing an overall OFF/ON/OFF switch effect during the time course of this experiment.

When dasatinib (100 nM) was added to the culture medium only on the second stimulation with CEA-TCB to prevent T cell cytotoxicity and then removed for the third stimulation, target cell killing was finally restored (figure 4D and online supplemental file 1), mimicking an ON/OFF/ON switch. Addition of 100 nM dasatinib for the second TCB stimulation prevented the release of IFN-γ, IL-2 and TNF-α, as expected (figure 4F).

Cytokine release has been shown to occur on initial exposure to T cell bispecific antibodies, but not or much less on subsequent dosing. Interestingly, low doses of dasatinib (<25 nM) added for the first TCB stimulation in the OFF/ON assay described in figure 4A partially inhibited target cell killing but seemed to equilibrate cytokine release between the first and second stimulation (online supplemental figure 10A–C).

In line with the previous observations with CEA-TCB and HLA-A2 WT1-TCB, 100 nM dasatinib prevented CD19-TCB-dependent killing of SU-DHL-8 tumor cells and T cell activation as well as the release of TNF-α, IFN-γ and IL-2 in vitro (figure 5A,B and online supplemental figure 11A,B). To verify whether dasatinib could prevent CD19-TCB-induced B cell depletion and cytokine release in vivo, humanized NSG mice were either treated with vehicle or with 0.5 mg/kg CD19-TCB as a monotherapy or combined with 50 mg/kg dasatinib (figure 5C). To best translate the clinical pharmacodynamics and pharmacokinetics profile of dasatinib and to verify if the resulting exposure would be sufficient to prevent CD19-TCB-induced T cell cytotoxicity and cytokine release, dasatinib was given orally three times on day 0 and twice on days 1 and 2. As shown by the CD20⁺ B cell count measured in the blood at 48 hours, dasatinib prevented the killing of CD20⁺ B cells by CD19-TCB (figure 5D). At 72 hours, partial killing of B cells was nevertheless observed (figure 5D). The half-life of dasatinib being around 6 hours, the exposure of dasatinib was probably not sufficient any longer to continuously inhibit T cell cytotoxicity at the later timepoint. This is another indication that the inhibitory effect of dasatinib is rapidly reversible in vivo, as observed by Mestermann et al29 and Weber et al30 for CAR T cells.33

Dasatinib blocked the release of IL-2, TNF-α, IFN-γ and IL-6 measured 1 hour 30 min and 6 hours after CD19-TCB treatment, indicating that dasatinib could also rapidly switch off CD19-TCB-dependent cytokine release (figure 5E,F).

In line with the in vitro findings, the rapid onset of the activity of dasatinib allowed to prevent B cell depletion and cytokine release induced by the first infusion of CD19-TCB in humanized NSG mice. Collectively, these data demonstrate the favorable pharmacodynamic profile of dasatinib, efficiently preventing CD19-TCB-induced T cell cytotoxicity and cytokine release for at least 48 hours when administered twice per day at the dose of 50 mg/kg.