Combination Therapy for Treating Advanced Drug-Resistant Acute Lymphoblastic Leukemia

Acute lymphoblastic leukemia (ALL) accounts for 80% of all leukemias in childhood, making it the most common pediatric leukemia. It also accounts for about 20% of adult leukemias (1, 2). Although CD19-mediated chimeric antigen receptor (CAR) T-cell immunotherapy has shown promise in B-ALL therapy, relapse due to loss of CD19 expression still remains as a challenge (3). Disease relapse and drug resistance are two major challenges in ALL therapy. Presence of Philadelphia chromosome (Ph) translocation, which encodes a constitutively active tyrosine kinase (4–8), and the tumor microenvironment are two major factors that play an important role in the development of ALL drug resistance (7, 8). Tumor cells themselves or/and other cells in the tumor microenvironment produce cytokines, chemokines, or growth factors which support the development and survival of drug-resistant cells. B-cell activating factor (BAFF) is one such factor, belonging to the TNF superfamily and is expressed by stromal cells, macrophages, and dendritic cells (9, 10). BAFF has three receptors, TACI, BCMA, and BAFF-R, out of which BAFF-R is the only specific receptor for BAFF (10). APRIL, a homologous ligand, also binds to TACI and BCMA (10, 11). BAFF-R is expressed mainly on mature B cells and mature B-cell malignancies such as chronic lymphocytic leukemia (CLL), multiple myeloma, and B-cell lymphomas (11). BAFF- and BAFF-R–deficient mice lack a mature B-cell compartment, demonstrating the involvement of BAFF signaling in survival of mature B cells (10–12).

We and others have previously shown that B-ALL cells express surface BAFF-R, making it an attractive antigen for therapeutic targeting (13, 14), and also report potential therapeutic approaches targeting BAFF-R, which include the fusion toxin rGel/BLyS and an antibody-dependent cellular cytotoxicity (ADCC) optimized anti–BAFF-R, B1239 (15–17). Antibody to a tumor-specific antigen can be used to target cancer cells through ADCC and is mediated mainly by CD16 (FcRγIII), a major triggering receptor on natural killer (NK) cells. There are many ADCC-optimized antibodies used in clinic, including anti-CD20 rituximab (Rituxan), anti–TNFα infliximab (Remicade), and anti-Her2 trastuzumab (Herceptin) (18–22). Rituximab is routinely used in therapy for treatment of adult B- ALL and CLL (21, 22). Monoclonal antibodies including ofatumumab, alemtuzumab, and epratuzumab are currently under investigation for B-ALL therapy (22–24).

Here, we showed that drug-resistant and relapsed ALL cells retain BAFF-R expression and that anti-BAFF-R mediates significant in vitro and in vivo NK cell–mediated ADCC against these cells. B1239, anti–BAFF-R, is renamed officially as VAY736. Because most of the in vivo preclinical studies use very early disease as a starting point for therapy, the effect of cytokines or growth factors produced by the tumor itself or tumor microenvironment on these treatment strategies are lacking. We showed that four injections of VAY736 lead to enhanced NK cell–mediated killing of ALL cells in vivo, provided the treatments are started early. If the treatments are started late, ADCC efficacy was inhibited by tumor microenvironmental factors. As reported previously by others (25), we confirmed that transforming growth factor-beta (TGFβ) was produced by ALL cells, and we demonstrated that TGFβ was a negative regulator of CD16 expression and, thus, NK cell ADCC function. After coincubation with ALL cells, NK cells had decreased CD16 expression, which could be reversed by EW-7197 (26), a TGFβ receptor 1 (TGFβR1) small-molecule inhibitor. A combination of EW-7197 and VAY736 significantly enhanced NK cell killing of ALL cells in vitro and in vivo. In summary, we demonstrated the in vivo therapeutic efficacy of the BAFF-R antibody VAY736 in combination with the TGFβR1 inhibitor EW-7197 against advanced-stage drug-resistant B-ALL disease.

Human peripheral blood mononuclear cells (PBMC) from healthy adult donors (12 females and 11 males) were obtained with informed consent at the Hematopoietic Stem Cell Core Facility (HSC), Case Western Reserve University. ALL patient blood and bone marrow samples were obtained from 5 pediatric and 11 adult patients from Dr. Rose Beck, University Hospitals, Cleveland and also from HSC core facility. Only discarded human blood and bone marrow samples were used in accordance with the common rule ethical standards, and informed consent for this study was approved by the University Hospital's Case Medical Center Institutional Review Board (IRB). Both male and female ALL patients as well as adults and pediatric ALL patients were included in this study. Blood and bone marrow samples were processed either for RBC lysis (Santa Cruz Biotechnology) or for Ficoll-paque gradient (GE Healthcare) following standard protocols. PBMCs were frozen in 95% FBS (Sigma) containing 5% DMSO (Fisher Scientific).

NK cells were expanded as previously described (27, 28). Ficoll-purified PBMCs (20 × 10⁶ cells) from healthy or ALL patients were cocultured with 10 × 10⁶ irradiated K562 clone 9 cell line expressing membrane-bound IL-21 for 2 weeks. K562 clone 9 cells were a kind gift from Dr. Dean A. Lee (Nationwide Children's Hospital, Columbus, OH). Cells were grown in RPMI-1640 medium supplemented with 10% FBS (Sigma) and 1% penicillin/streptomycin (HyClone). Fresh media were added every third day containing IL2 (100 U/mL; PeproTech). NK cells were purified at the end of the 2-week coculture using the MojoSort Human NK cell Isolation Kit (BioLegend). NK cell purity (>95%) was determined via flow cytometry (BD Accuri C6) using CD56 (5.11H11) and CD3 (HIT3a) antibodies from BioLegend.

NOD/SCID/IL2rg^−/− (NSG) mice were purchased from Case Western Reserve University (CWRU) Athymic Animal Core Facility (Cleveland, OH). Mice were maintained in pathogen-free conditions at the CWRU animal facility. All animal experiments were performed in accordance with and with the approval of CWRU's Institutional Animal Care and Use Committee (IACUC) and NIH guidelines.

NSG mice were used to expand ALL patients (PT) cells: PT-1 (newly diagnosed), PT-2 (relapse), PT-3 (drug resistant), and PT-4 (drug resistant), which were further used in the experiments (described below). For expansion purposes, NSG mice were transplanted intravenously with 2 × 10⁶ patient-derived ALL cells. ALL cells were allowed to proliferate, and mice were sacrificed when terminally ill: signs of hind limb paralysis, splenomegaly which appears around days 35 to 42, or if mice lose more than 20% of its body weight. Splenic cells from mice were cultured overnight in MEMα (Sigma) media supplemented with 20% FBS (Sigma), 1% L-glutamine (Gibco), and 1% penicillin/streptomycin (HyClone), after which nonadherent cells were analyzed by flow cytometry using the Accuri C6. Cells >95% CD19⁺ (4G7) and CD10⁺ (H110a) were frozen in 95% FBS, 5% DMSO (Fisher Scientific) in aliquots containing 10 million cells for further use.

For in vivo experiments, NSG mice were transplanted intravenously with 2 × 10⁶ patient-derived drug-resistant ALL cells. ALL cells were allowed to proliferate in vivo for the indicated time prior to treatments. Transplanted mice (n = 3–6 per group) were either injected i.p. with VAY736 (10 mg/kg; NCT02149420), PBS (control mice), or oral administration of EW-7197 (3 mg/kg; ref. 26). Expanded NK cells (expanded for 2 weeks) were injected i.v. to either the VAY736- or EW-7197-treated groups or the NK cell–treated-only group at the indicated time points. Mice were sacrificed at the indicated times. Bone marrow, spleen, and blood were analyzed by flow cytometry for the presence of human ALL cells using human CD19 and CD10 antibodies, as described below. The survival effect of VAY736 + EW-7197 combination treatment compared with control (PBS), VAY736, or EW-7197 single treatments was determined by Kaplan–Meier analysis and log-rank test. All statistical analyses were performed using GraphPad Prism (GraphPad Software Inc.).

Blood was collected with a 2-mL syringe from the heart of mice into PBS-containing tubes followed by Red Blood Cell Lysis (Santa Cruz Biotechnologies) using the standard protocol. Bone marrow was collected by flushing femurs and tibias with PBS containing 5% FBS. Spleens were smashed with syringe plunges in PBS containing 5% FBS. Bone marrow and spleen suspensions were filtered through a 70-μm nylon filter. Bone marrow and splenic cells (1 × 10⁶) were aliquoted for extracellular staining in 100 μL volume of PBS containing 5% FBS. Live-cell counts were determined by excluding Trypan blue-positive cells. Data collection and analysis was performed using the Accuri C6 flow cytometer and software (BD Biosciences).

For expression of BAFF-R in healthy donor immune cells, nonexpanded PBMCs (1 × 10⁶ per stain) were washed and stained with CD3 (HIT3a), CD19 (4G7), CD56 (5.11h11), IgG (mopc-21), BAFF-R (11c1) from BioLegend. To assess NK cell receptor expression, 1 × 10⁶ NK cells from the indicated donors (expanded for 2 weeks) were washed and stained with CD56 (5.11h11), CD3 (HIT3a), NKp46 (9E2), NKp30 (p30-15), NKG2D (ID11), and KLRG1 (14C2A07). To analyze the expression of HLAE and MICA/B, the indicated ALL cells were thawed overnight in MEMα (Sigma) media supplemented with 20% FBS (Sigma), 1% L-glutamine (Gibco), and 1% penicillin/streptomycin (HyClone). Then, 1 × 10⁶ indicated ALL cells were washed and stained with CD19 (4G7), CD10 (H110a), HLAE (3D12), and MICA/B (6D4) from BioLegend. For the analysis of CD69 expression, indicated expanded NK cells (1 × 10⁶) were cocultured with (2 × 10⁵) PT-3 ALL cells that had been pretreated with or without VAY736 (20 μg/mL for 2 hours) for 4 hours. Cocultures were then washed and stained with CD56 (5.11h11), CD3 (HIT3a), and CD69 (FN50) from BioLegend. All data collection was done on Accuri C6 flow cytometer, and data analysis was performed using Accuri C6 software (BD Biosciences).

The human anti-BAFF-R was obtained from Novartis. EW-719 is a synthetic small-molecule inhibitor of TGFβR1 purchased from Cayman Chemicals. Flow cytometry antibodies for BAFF-R (11c1), IgG (mopc-21), CD56 (5.11h11), CD3 (HIT3a), CD19 (4G7), CD10 (H110a), CD16 (3G8), NKp46 (9E2), NKp30 (p30-15), NKG2D (ID11), KLRG1 (14C2A07), HLAE (3D12), MICA/B (6D4), and CD69 (FN50) were purchased from BioLegend. Secondary anti-human IgG was purchased from Invitrogen (62-8411). Western blot antibodies p100 and p50 (4882), A-1 (14093), and BCL-XL (2764) were purchased from Cell Signaling, and Pim-1 (13513) and GAPDH (47724) were purchased from Santa Cruz Biotechnology. Recombinant TGFβ (580702) and BAFF (591202) were purchased from BioLegend.

OP9 bone marrow stroma (ATCC), K562 clone 9 chronic myelogenous leukemia (Dean A. Lee), Jurkat T lymphocytes (Parameswaran Ramakrishna, CWRU, Cleveland, OH), Jeko-1 mantle cell lymphoma (ATCC) were passaged >10 times. Briefly, OP9 cells were cultured in MEMα supplemented with 20% FBS (Sigma), 1% penicillin/streptomycin (HyClone), 1% L-glutamine (Gibco), whereas the other cell lines were cultured in RPMI (Sigma) media supplemented with 10% FBS (Sigma), 1% L-glutamine (Gibco), and 1% penicillin/streptomycin (HyClone). Mycoplasma detection was performed using the Lonza Mycoalert Mycoplasma Detection Kit, but cell line authentication was not routinely assessed.

Expanded ALL cells (1 × 10⁶, >95% CD19⁺, CD10⁺) were incubated with BAFF-R antibody (5 μg/mL) for 30 minutes at room temperature, washed with PBS, and incubated with anti-human IgG diluted 1:50 (Invitrogen), followed by fluorescence-activated cell sorting (Accuri C6). For competition experiments, ALL cells were incubated with BAFF (0.02, 0.05, 1, or 3 μg/mL), followed by the addition of BAFF-R antibody (5 μg/mL) for 30 minutes at room temperature. Binding of BAFF-R antibody to the BAFF-R on ALL cells was determined via flow cytometry (Accuri C6).

Serum from mice transplanted with PBS or PT-3 ALL cells was obtained by centrifugation of 500 μL of blood (10,000 × g for 10 minutes) on day 35. Serum samples were stored at −80°C and thawed prior to use. Mouse TGFβ was measured with the Legend Max Mouse Latent TGFβ ELISA Kit (BioLegend) and 1:50 serum dilution in manufacturer's buffer was performed as specified by the manufacturer's instructions. Per condition, 4 μL of sample was diluted with 196 μL of Assay Buffer provided by the manufacturer, after which 50 μL was added to the well in triplicates. The Human/Mouse TGF-beta 1 Uncoated Elisa Kit (Invitrogen) was also used to measure TGFβ, and serum was diluted 1:5 in PBS, following specific instructions from the manufacturer. Per condition, 20 μL of serum was diluted in 80 μL of PBS (1:5). Samples were activated with acid: 20 μL of 1N HCL (Fisher Scientific) was added to 100 μL of serum diluted in 80 μL of PBS, incubated at room temperature for 10 minutes followed by the addition of 20 μL of 1N NaOH, after which 100 μL was added to each well in triplicates. Correction to the dilution factor of 1.4 was made for final calculations. For granzyme B, TNFα, and IFNγ ELISAs, indicated NK cells (1 × 10⁶) were cocultured with (2 × 10⁵) PT-3 ALL cells that had been pretreated with or without VAY736 (20 μg/mL for 2 hours) for 4 hours. Legend Max Human Granzyme B ELISA Kit was obtained from BioLegend. TNFα and IFNγ ELISAs were obtained from Invitrogen. Per condition, 50 μL (granzyme B ELISA) and 100 μL (TNFα, and IFNγ ELISAs) of nondiluted sample was added to each well in triplicates. Binding of TGFβ, granzyme B, TNFα, and IFNγ was detected using secondary antibody, streptavidin-HRP, and TMB substrate solution (provided with specified ELISA kit). Substrate conversion was stopped after 20 minutes with 100 μL stop solution (1M H₃PO₄) provided with the ELISA Kits. Plates were washed with PBS plus 0.05% Tween20 in-between incubations. Assay diluent provided by the manufacturer or RPMI medium (Sigma) was used as negative controls, and specific standard proteins were used as positive controls. Standard reconstitutions and curves were generated as per manufacturer's instructions for each assay. Optical density values were obtained using a microplate reader set to 450 nm (Bio-Rad iMark Microplate reader). The derived TGFβ, granzyme B, TNFα, and IFNγ concentrations (ng/mL) were determined using specific standard curve-derived formulas. The derived TGFβ concentrations (ng/mL) were multiplied by 5 to correct for dilution.

ALL cell killing by NK cells was analyzed using calcein-AM assay purchased from Life Technologies (17). Target tumor cells (10 × 10⁶) were labeled with (0.5 μmol/L) calcein-AM for 30 minutes at 37°C. Following staining, cells were washed with PBS, counted using Trypan blue (Sigma), and 2 × 10⁶/mL cells were incubated with VAY736 (20 μg/mL) for 2 hours. After washing 2 times with PBS, ALL (10 × 10³) cells per well (96-well plate) were cocultured at 37°C with MACs sorted indicated NK cells (5 × 10⁴) at 5:1 ratio for 4 hours. NK cell purity was >95% as determined by CD56⁺, CD3⁻ cells via flow cytometry. The percentage of specific lysis from triplicate wells was determined using the following equation, in which “AFU mean spontaneous release” is calcein-AM release by target cells in the absence of antibody or NK cells and “AFU mean maximal release” is calcein-AM release by target cells upon lysis by detergent. % specific lysis = 100 × (AFU mean experimental release−AFU mean spontaneous release)/(AFU means maximal release−AFU means spontaneous release).

The indicated ALL cells were plated at a density of 1 × 10⁶ cells per mL in different conditions in an irradiated OP9 bone marrow stroma cells (16, 17). OP9 cells were purchased from ATCC. Coculture of human ALL cells with OP9 cells performed in MEM-α medium supplemented with 20% FBS (Sigma), 1% L-glutamine (Gibco), 1% penicillin/streptomycin (HyClone). The rate of proliferation and viability was monitored every other day by manually counting the viable cells and total cells using Trypan blue stain (Sigma). For the E670-based proliferation assays, 5 × 10⁶ ALL cells were labeled with 1 μm E670 (Invitrogen) and cultured for 7 days. E670 dilution was measured by gating on live E670⁺ cells using flow cytometry (Accuri C6).

For testing drug resistance in vitro, 1 × 10⁶ cells (PT-1, PT-2, or PT-3) per mL were cultured in different conditions in an irradiated OP9 bone marrow stroma cells. Vincristine (2.5 nmol/L) or nilotinib (300 nmol/L) were added every other day. Nonadherent leukemia cells were carefully collected from the stromal layer. Viability of the ALL cells was determined by excluding Trypan blue–positive ALL cells.

A flow-based Annexin V/propidium iodide (PI) assay (BioLegend) was used to measure ALL cells apoptosis following the vendor's protocol. Briefly, 1 × 10⁶ tumor cells were washed in PBS, resuspended in 100 μL Annexin V–binding buffer (BioLegend), and stained with 1 μg/mL APC-conjugated Annexin V (BioLegend) for 15 minutes on ice in the dark. Cells were washed and stained with PI for 5 minutes at room temperature in the dark. Tumor cell apoptosis was evaluated by gating on Annexin V and PI double-positive cells via flow cytometry (Accuri C6).

Data and corresponding statistics were downloaded directly from https://www.oncomine.org. BAFF-R mRNA expression analysis from 359 pediatric, 147 adults ALL patients, and 74 healthy donors analyzed on Affymetrix U133 plus microarray was obtained directly from the Haferlach Leukemia Dataset (Reporter ID: 1552892_at). TGFβ mRNA expression from 20 B-cell ALL and 8 healthy lymphocyte samples analyzed on Affymetrix U133A microarray was downloaded from the Mia Leukemia Dataset (Reporter ID: 203085_s_at).

Five million expanded NK cells were labeled with eFluor 670 (APC; eBioscience), and 5 × 10⁶ ALL cells were labeled with CellTracker Green CMFDA (FITC; Invitrogen), according to the manufacturer's protocol. Cells were coincubated at 37°C for 30 minutes at a ratio of 2:1 (NK:tumor). The proportion of double-positive cells (conjugate formation) was analyzed using BD Acuri C6 flow cytometer.

Total RNA was isolated from 2 × 10⁶ of the indicated cells using High pure RNA isolation kit from Roche Life Science. RNA concentration and quality were determined by a Thermo Fisher Nanodrop Lite Spectrophotometer.

cDNA synthesis was performed on 1 μg RNA per sample using the Bio-Rad iScript Reverse Transcription Supermix cDNA synthesis kit. qRT-PCR was performed in triplicates using the Bio-Rad iQ SYBR Green Supermix according to the manufacturer's instructions. GAPDH was used for normalization, and BAFF-R relative mRNA expression was determined using the ΔΔCT method. The following primers were used:

• BAFF-R: Forward CCCTGGACAAGGTCATCATT

• BAFF-R: Reverse TCTTGGTGGTCACCAGTTCA

• GAPDH: Forward TGGAAATCCCATCACCATCTT

• GAPDH: Reverse CCTGCTTCACCACCTTCTT

ALL PT-3 cells (2 × 10⁶) were cultured in MEMα with either VAY736 (20 μg/mL), recombinant BAFF (200 ng/mL), or combination for 72 hours. Indicated cells were washed with PBS, followed by lysis with 9 M urea buffer (Fisher Chemicals) in Tris Buffer (Fisher Bioreagents). Supernatant was collected after centrifuging lysed cells at 14,000 × g for 10 minutes at 4°C, and protein concentration was estimated using Thermo Scientific Nanodrop Lite Spectrophotometer. Protein samples (30 μg) were run on an SDS-polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane (Bio-Rad). The membrane was blocked for 1 hour with 5% nonfat dry milk prepared in Tris Buffer Saline containing 0.1% Tween 20 (Fisher Scientific). Membranes were probed with the primary antibodies overnight at 4°C, and with secondary antibodies for 1 hour at room temperature. Respective mouse or rabbit HRP-conjugated secondary antibodies were used to detect specific protein bands. Blots were developed using Clarity Western ECL substrate (Bio-Rad).

Applicable data were analyzed using unpaired Student t test using Prism 8 software. All in vitro experiments were done in triplicates. P values assigned were ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. Statistical analysis of mice survival curve was done by log-rank (Mantel–Cox) test using GraphPad Prism software Inc.

BAFF-R is known to be aberrantly expressed on B-ALL (13, 14). To confirm BAFF-R expression in ALL cells, we analyzed 359 pediatric and 147 adults ALL patient samples compared with PBMCs from 74 healthy donors using the Haferlach Leukemia Oncomine database (29). In accordance with previous reports, we detected a significant fold increase in BAFF-R expression in pediatric (1.3, P = 1.08 × 10⁻¹⁹) and adult ALL cells (1.4, P = 1.6 × 10⁻²²) compared with PBMCs (Fig. 1A). We also detected variable BAFF-R expression in 11 adults and 5 pediatric ALL patient blasts by flow cytometry (Fig. 1B; Table 1). mRNA expression was also analyzed by RT-PCR (Supplementary Fig. S1A). B lymphocytes (CD19⁺CD10⁻) from normal donors also expressed BAFF-R but were absent on NK or T cells (Supplementary Fig. S1B). Because drug resistance and relapse are two major problems in ALL, we tested whether BAFF-R expression was maintained in relapse and drug-resistant patient ALL cells. Drug-resistant patient cells developed resistance to vincristine (ph⁻) or tyrosine kinase inhibitor nilotinib (ph+; Supplementary Fig. S1C) in a coculture OP9 stromal system (16, 17). Both relapse and drug-resistant ALL cells expressed surface BAFF-R (Fig. 1C; Table 1). We next tested the binding efficacy of VAY736 (anti-human BAFF-R optimized for ADCC) to drug-resistant and relapsed ALL cells. VAY736 was able to bind to newly diagnosed, drug-resistant, and relapsed ALL cells, as detected with human IgG (Fig. 1D and E). We previously reported that this antibody is not internalized upon binding to BAFF-R in ALL cells (17). Because surface presentation of antibody is a crucial factor in determining ADCC efficacy, we tested how long VAY736 remains bound to the surface of ALL cells. VAY736 can be detected on ALL cell surface even after 48 hours, making it ideal for ADCC (Fig. 1F). Normal serum BAFF concentrations are in the range of 0.3–2 ng/mL, and it is elevated in autoimmune diseases such as SLE to a range of 0.5–6 ng/mL (30). We speculated that endogenous BAFF would compete with VAY736 binding to BAFF-R on ALL cells. We preincubated ALL cells with human recombinant BAFF for 2 hours, followed by the addition of VAY736. Preincubation of ALL cells with 50 ng/mL of human recombinant BAFF did not affect binding of VAY736. BAFF concentrations (1 or 3 μg/mL), which is higher than endogenous BAFF, competed with VAY736 binding (Fig. 1G).

To address the effect of VAY736 on the proliferation of ALL cells, we monitored ALL cell proliferation for a week in the presence or absence of VAY736 (20 μg/mL). Proliferation of VAY736-treated ALL cells was comparable with untreated ALL cells, and addition of recombinant BAFF or a combination of VAY736 and recombinant BAFF did not affect ALL cell proliferation (Fig. 2A). We also analyzed viable cell counts and percent viability using Trypan blue in PT-3 and PT-4 ALL cells treated with BAFF, VAY736, and BAFF + VAY736 compared with nontreated control ALL cells. No significant difference in viable cell count or percent viability was seen in VAY736-treated ALL cells compared with other groups, even after 6 days of incubation (Fig. 2B and C). There was no noticeable cell death in VAY736-, BAFF-, or VAY736 + BAFF-treated PT-3 and PT-4 ALL cultures (Fig. 2D). BAFF is known to induce alternate NF-kB signaling in normal B cells (31, 32) without affecting their proliferation or survival rate. As expected, BAFF induced alternate NF-kB signaling in ALL cells, as evidenced by p100 to p52 conversion (Supplementary Fig. S2), and addition of VAY736 partially inhibited this. VAY736 treatment alone did not significantly change p52 levels. BAFF-induced survival proteins also behaved similarly, without showing any significant changes in VAY736 alone-treated cells, but inhibited in BAFF + VAY736-treated cells compared with BAFF alone–treated cells (Supplementary Fig. S2).

We next tested the efficacy of VAY736 to mediate NK cell–mediated ADCC of ALL cells. Compared with NK cells alone (3%–40%), VAY736 significantly increased NK cell–mediated cytotoxicity of drug-resistant and relapse ALL cells (15%–70%) depending on the NK cell donor (Fig. 3A). ALL cells from four different patients (PT-1, PT-2, PT-3, and PT-4) and NK cells from 3 different donors (NK-2, NK-3, and NK-4) were used to confirm the effect of VAY736 in mediating ADCC. NK cell receptor expression and the corresponding ligand expression by tumor cells contribute to NK cell heterogeneity in tumor cell killing. We analyzed expression of receptors, such as NKP46, NKp30, NKG2D, KLRG1 on NK cells (Supplementary Fig. S3A), and MICA/B, an HLA-E ligand on ALL cells (Supplementary Fig. S3B), and found that receptor expression (Supplementary Fig. S3A) was donor specific for NK cells, whereas the ligands did not show much variability (Supplementary Fig. S3B). The cytotoxic function of NK cells was tested after optimizing the effector (NK) to target (ALL) ratio to be 5:1 (Supplementary Fig. S3C) for in vitro ADCC experiments. ALL cell killing by NK cells varied from donor to donor (Supplementary Fig. S3D).

NK cells from cancer patients are reported as dysfunctional (25, 33–35). Hence, we asked whether the patients' own NK cells could mediate ADCC in the presence of VAY736. We tested the ADCC-promoting activity of VAY736 on NK cells expanded from healthy donors (allogeneic) and ALL patients (autologous). VAY736 mediated enhanced killing of ALL cells in the presence of both allogeneic and autologous NK cells compared with NK cells alone (Fig. 3B). ALL cells from 2 different patients, PT-5 and PT-7, were used in this assay.

We then analyzed NK-ALL conjugate formation and found that more conjugates are formed in VAY736 + NK + ALL cultures compared with NK + ALL cell cultures (Fig. 3C). CD69 expression on NK cells was also significantly increased in the presence of VAY736, as confirmed using NK cells from 4 different patients (NK-6, NK-8, NK-9, and NK-10; Fig. 3D). Secretion of TNFα, IFNγ, and granzyme B, major players in NK cell cytotoxic function were significantly increased in the presence of VAY736 (Fig. 3E).

We transplanted NSG mice with drug-resistant ALL (2 × 10⁶) cells and allowed them to proliferate for 3 days to mimic early-stage disease. Mice were treated with either 10 mg/kg VAY736 with or without 3 × 10⁶ NK cells, NK cells alone, or PBS on days 3, 10, 17, and 24 (Fig. 4A). On day 31, mice were sacrificed, and tumor burden in the bone marrow and spleen was assessed by the percentage of CD19⁺CD10⁺ human cells in the live-cell gate (Supplementary Fig. S4A). Leukemia burden in the bone marrow (∼80%) and spleen (∼90%) of control untreated mice was analyzed as shown in Fig. 4B. VAY736 alone (∼55%) and NK cell–treated (∼60%) mice showed reduced tumor burden in the bone marrow, but tumor burden in the spleen was similar to control mice. This is consistent with our in vitro data (Fig. 2B and C) showing little effect of VAY736 on cell viability or proliferation of ALL cells. However, a significant reduction of tumor load was seen both in spleen (40%) and bone marrow (27%) of the VAY736 + NK combination–treated group (Fig. 4B and C). Spleen weights and sizes (Supplementary Fig. S4B) of VAY736 + NK-treated mice were reduced by ∼50% compared with controls or NK alone–treated mice. It was indeed surprising to notice the lack of ALL killing in spleen of mice treated with NK cells alone, despite similar percentages of NK cells (CD56⁺CD3⁻) detected in the spleen or bone marrow of NK- and NK + VAY736-treated mice (Supplementary Fig. S4C). Survival of leukemic mice treated with VAY736 and NK cells was also significantly enhanced compared with VAY736 alone, NK alone, or PBS-treated mice (Fig. 4D and E).

Because drug-resistant and relapsed patients do not respond well to conventional treatment strategies, they often develop advanced-stage disease. Figure 4F shows the leukemic burden in mice on day 5 (early) and day 15 (advanced). We further investigated whether VAY736 + NK cell treatment was effective in an advance leukemia disease model. We started treatment on day 14 and four treatments were performed on days 14, 21, 28, and 35. On day 42, we analyzed tumor burden (Fig. 4G). VAY736 + NK treatment significantly reduced disease burden in the spleen and bone marrow of leukemic mice compared with control mice (Fig. 4H). However, NK cell killing of drug-resistant ALL cells was not as pronounced, as was seen in our previous experiment, where treatment was started early (day 3). Together, these data suggest that the ALL tumor microenvironment renders NK + VAY736 treatment less effective.

NK cells from ALL patients exhibit impaired cytotoxic function (25), and factors in the tumor microenvironment contribute to decreased expression of NK cell–activating receptors (25, 36). One such factor is TGFβ, a negative regulator of NK cell cytotoxic function in ALL patients (25). To confirm TGFβ expression in ALL cells, we compared TGFβ mRNA expression in pre-B cells from 5 pediatric healthy donors compared with 18 pediatric B-ALL patient samples using the Maia leukemia database on Oncomine (37). TGFβ expression was increased in B-ALL compared with normal counterpart pre-B cells (Fig. 5A). We further confirmed this by analyzing serum TGFβ from ALL or PBS-injected mice. Human TGFβ was enhanced in the serum of leukemic mice compared with control mice (Fig. 5B), whereas mouse TGFβ did not show significant changes (Supplementary Fig. S5A). CD16 expression on NK cells is essential for ADCC. TGFβ reduced the percentage of CD16⁺ NK cells from 28% to 12% (Fig. 5C) and reduced NK cell cytotoxicity against ALL cells from 30% to 17% (Fig. 5D). Addition of EW-7197, a potent TGFβRI inhibitor, to the culture restored CD16 expression on NK cells and NK cell cytotoxicity against ALL cells (Fig. 5C and D). Coculturing NK cells with drug-resistant ALL cells also led to decreased expression of CD16 on NK cells, and this effect was partially ameliorated by EW-7197 (Fig. 5E). We next tested whether the combination of VAY736 and EW-7197 would further increase NK cell killing of anti-BAFF-R–targeted ALL cells. The combination of VAY736 and EW-7197 further increased NK cell lysis of ALL cells by 10% to 35% using 4 different ALL patient cells (Fig. 5F).

EW-7197 has been reported to exert therapeutic activity against solid tumors (NCT02160106, NIH). We first examined whether EW‐7197, which inhibits intracellular TGFβ signaling, has antitumor activity against ALL by treating NSG mice bearing drug-resistant ALL cells with oral EW-7197 three times per week starting from day 14. On day 42, we analyzed tumor burden (Fig. 6A). A comparable tumor load in the spleen and bone marrow of EW-7197–treated mice was seen compared with control untreated leukemic mice (Fig. 6B). To evaluate the in vivo efficacy of combination EW-7197, NK cells, and VAY736, we treated NSG mice bearing drug-resistant PT-3 cells with daily oral administration of EW-7197 alone or in combination with weekly VAY736 and NK cells or PBS (Fig. 6C). At day 35, control mice were terminally ill, and we sacrificed all mice to analyze tumor burden in the spleen, bone marrow, and blood. Leukemia cell frequency was significantly reduced in the bone marrow (29%–36%), spleen (17%–20%), and blood (10%–13%) of EW-7197 + NK-treated mice compared with untreated control mice (Fig. 6D). Leukemia cell frequencies were significantly reduced in the bone marrow (36%–44%), spleen (19%–24%), and blood (43%–50%) of VAY736 + NK-treated mice compared with control PBS-treated mice. However, with the combination EW-7197 + VAY736 + NK treatment, there was a significant reduction in ALL cell growth in the bone marrow (50%–58%), spleen (46%–48%), and blood (62%–77%) compared with control PBS-treated mice. The combination of EW-7197 + VAY736 + NK cells showed significantly reduced percentage of ALL cells in the bone marrow, spleen, and blood compared with NK + EW-7197 or NK + VAY736-treated groups (Fig. 6E). Spleen weights and sizes from the mice treated with combination of EW-7197 + VAY736 + NK cells were significantly reduced compared with control untreated mice (Supplementary Fig. S5B). Similar to the in vitro findings in Fig. 5C–E, NK cells from EW-7197–treated mice showed increase in CD16 expression compared with NK + VAY736–treated mice (Supplementary Fig. S6A). We performed another in vivo experiment by injecting PT-4 ALL cells to confirm the effect of EW-7197 to enhance NK cell–mediated ADCC and similar results were obtained (Fig. 6F–H; Supplementary Fig. S6B). Because cancer cells tend to lose target antigens and develop resistance to targeted therapies, we analyzed BAFF-R surface expression in PT-4 ALL cells from control PBS-injected and EW-7197 + VAY736 + NK cell combination–treated mice at day 42. We found that BAFF-R expression was not affected after VAY736 targeting (Fig. 6I). These data indicated that TGFβ produced by ALL cells and the tumor microenvironment negatively influenced NK cell–mediated ADCC, and by inhibiting the TGFβ receptor on NK cells, we could partially reverse this dysfunction (Supplementary Fig. S6C).

Adoptive cell therapy using NK cells is of high priority for leukemia treatment, but often NK cells are dysfunctional in these patients or tumor cells escape from NK cell–mediated killing by various mechanisms (36). ALL cells have downregulated expression of MICA and MICB, which are ligands for activating receptors on NK cells, whereas inhibitory ligands such as HLA class 1 have no change, which makes NK cell therapy difficult in these patients (36). Here, we showed that NK cells in the presence of VAY736 exhibits enhanced ALL cell killing by ADCC. ADCC is a rapid process mediated by effector immune cells like NK cells, macrophages, and neutrophils (38, 39). We assessed the binding affinity of VAY736 to drug-resistant ALL cells and found that the antibody was surface bound as early as 30 minutes and still remained on the surface after 48 hours. This demonstrated that VAY736 is presented on the cell surface of ALL cells, improving NK cell–mediated ADCC efficacy. BAFF-R expression on surface of ALL cells varies from patient to patient, hence confirming BAFF-R expression on ALL blast cells is a prerequisite for treatment with VAY736 antibody. Our data point to reduced BAFF-R expression in pediatric ALL patients compared with adult ALL patients, but 16 ALL patients are not sufficient to draw a conclusion and need further analysis using larger patient data sets. Another factor that could possibly affect binding of VAY736 to ALL cells is competition with serum BAFF in patients (17). Increased BAFF expression is reported in ALL patients (40), so we evaluated the competition of VAY736 with BAFF. Our data showed no competition of VAY736 binding when exposed to endogenous BAFF. BAFF is a survival factor for mature B cells and inhibiting BAFF signaling affects mature B-cell antigen-induced proliferation (9, 17, 31, 32). VAY736 did not inhibit BAFF-induced signaling pathways or survival genes in ALL cells at basal state, which could explain why VAY736 monotherapy did not show any effect on survival or proliferation of ALL cells. Also, VAY736-induced ALL cell killing is solely via ADCC. Addition of exogenous BAFF also had no effect on ALL cell basal survival or proliferation, similar to normal B cells (31).

NK cells from different donors exhibited a different degree of ALL cytotoxic activity, and this could be explained by heterogeneous nature of human NK cells. Tumor killing function of NK cells depends on various factors, including the nature of the NK cell subset, activating/inhibitory receptor repertoire, and interaction of HLA on tumor cells and KIRs on NK cells (36). VAY736 and NK cell combination treatment in vivo also killed ALL cells in PDX-ALL mouse models, provided the treatment started early before establishment of a significant tumor load or suppression in the tumor microenvironment. Delayed treatment onset was not as efficient in killing ALL cells both in the bone marrow and spleen. In real-life situations, most patients have a high tumor load and a suppressive tumor microenvironment at disease presentation, especially in drug-resistant and relapsed patients who fail to respond to conventional therapies. Because of this, it is important to evaluate the efficacy of novel treatment strategies in an advanced-stage tumor model in preclinical drug testing.

TGFβ is reported to inhibit interleukin-15 (IL15)-induced activation of mTOR and IFNγ production in NK cells, downregulates NKp30 and NKG2D (activating NK receptors), inhibits CD16-associated γ-chain, resulting in decreased surface expression of CD16, and thus contributing to NK cell dysfunction (41–44). TGFβ is produced by both tumor cells and NK cells in the tumor microenvironment. Hence, neutralizing TGFβ signaling in NK cells could contribute to enhanced ADCC performance. We observed that exposure of NK cells to TGFβ leads to suppression of ADCC via CD16 downregulation, whereas inhibiting TGFβR1 on ex vivo–expanded NK cells using EW-7197 partially reversed this ADCC inhibition. EW‐7197 has already been certified as an Investigational New Drug, and phase I clinical trials are ongoing (ref. 45; NCT02160106, NIH). Similarly, VAY736 is also used in clinical trials for various diseases including pemphigus vulgaris (NCT01930175, Novartis), rheumatoid arthritis (NCT02675803, NIH), Sjogren syndrome (NCT0296289, NIH), idiopathic pulmonary fibrosis (NCT03287414), autoimmune hepatitis (NCT03217422), and CLL (NCT03400176, NIH), making the combination therapy more feasible in ALL patients.

Because TGFβR1 inhibitors TEW-7197 (45) and LY2157299 (46) are used in clinical trials for solid tumors, as well as refractory multiple myeloma (NCT03143985), we analyzed the effect of EW-7197 on ALL cells in vivo. No significant change in tumor load was seen after EW-7197 treatment. Hence, increased ALL cell killing in the presence of EW-7197 and NK cells will be due to inhibition of TGFβ signaling in NK cells together with the increased numbers of CD16⁺ NK cells. VAY736 can also kill tumor cells through antibody-mediated phagocytosis (ADCP), as macrophages can be a major effector cell type that mediates cell killing through antibodies (38, 39). In this study, we injected only human NK cells for performing ADCC, and we assume injecting macrophages will further increase ALL killing by ADCP. In patients, both NK cells and macrophages can mediate ADCC or ADCP, thus more likely to eradicate B-ALL cells. Another way to enhance ADCC is to use TGFβ inhibitors in combination with EW-7197 to inhibit TGFβ signaling in NK cells, but this combination has to be further tested.

In summary, we showed that early treatment with VAY736 enhances ADCC in BAFF-R⁺ drug-resistant PDX-ALL mice, but if advanced disease is present, microenvironmental factors, such as TGFβ, could inhibit ADCC efficacy. Therefore, inhibition of TGFβ signaling in NK cells is required to achieve a significant therapeutic effect. We showed here that a combination of EW-7197 and VAY736 enhanced NK cell–mediated ADCC killing of BAFF-R⁺ ALL cells, even at advanced stages of disease or difficult-to-treat drug-resistant cells.

H. Gram is director of Novartis Pharma AG. No potential conflicts of interest were disclosed by the other authors.

Conception and design: R. Parameswaran, Y. Vicioso, R. Beck, J. Letterio

Development of methodology: Y. Vicioso, H. Gram

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Parameswaran, Y. Vicioso, R. Beck, A. Asthana, K. Zhang, D.P. Wong

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Parameswaran, Y. Vicioso, R. Beck, A. Asthana, K. Zhang, J. Letterio

Writing, review, and/or revision of the manuscript: R. Parameswaran, Y. Vicioso, H. Gram, R. Beck, A. Asthana, D.P. Wong, J. Letterio

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Parameswaran, Y. Vicioso, J. Letterio

Study supervision: R. Parameswaran

Other (provision of critical reagent): H. Gram

This work was supported by the Athymic Animal, Hematopoietic Biorepository and Cellular Therapy, Radiation Core facilities, Flow Cytometry and other Common Resources of the Case Comprehensive Cancer Center. This work was supported by NIH 1R21CA201775-01A1 (R. Parameswaran), St. Baldrick's Foundation (R. Parameswaran), The Andrew McDonough B+ Foundation (R. Parameswaran), and The Jane & Lee Seidman Chair in Pediatric Cancer Innovation (J. Letterio).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.