TIM-3 blockade enhances ex vivo stimulated allogeneic NK cell therapy for relapsed murine neuroblastoma after hematopoietic cell transplant

Abstract

Background

High-risk neuroblastoma (HR-NBL) is an aggressive tumor of the sympathetic nervous system with high risk of relapse and poor overall survival. Allogeneic hematopoietic cell transplant (allo-HCT) has been used previously in patients with HR-NBL; however, graft-versus-host disease (GVHD) and disease progression have limited clinical application. Ex vivo stimulated allogeneic natural killer (NK) cells represent an approach to enhance the graft-versus-tumor (GVT) effect without exacerbation of GVHD but have not shown efficacy in NBL.

Methods

Ex vivo stimulated NK cells from C57BL/6NCr (B6) mice were expanded with soluble interleukin-15 (IL-15) and IL-15 receptor alpha (IL-15Rα) alone or with irradiated CD137L/CD54⁺ aggressive variant of the Neuro-2a murine neuroblastoma cell line (15–4P) at a 1:1 ratio for 10–12 days. Allogeneic NK cells were then analyzed for activation, proliferation, cytokine production, and cytotoxicity against two murine NBL cell lines, Neuro2a and NXS2, in the absence or presence of anti-T-cell immunoglobulin and mucin-domain containing-3 (TIM-3). Lethally irradiated B6AJF1 mice received allo-HCT from B6 donors followed by NBL challenge after 7 days to mimic tumor relapse. Select groups received anti-TIM-3 starting on day 9 for every 4 days with/without infusions of 15–4P B6 NK cells on days 14, 21, and 28. In select experiments, T cell and NK cells were selectively depleted to establish contribution to the GVT effect. All groups were analyzed for tumor growth, GVHD and survival.

Results

Co-culturing NK cells with 15–4P results in 78-fold expansion with increased expression of Kiel-67 (Ki-67) and Natural Killer Group 2, Member D (NKG2D), NKp46, TNF-Related Apoptosis-Inducing Ligand (TRAIL) and TIM-3. 15–4P stimulated allogeneic NK cells showed enhanced cytotoxicity against NBL compared with IL-15 NK cells alone but was limited in part due to high expression of TIM-3 ligands on Neuro-2a compared with NXS2. The addition of TIM-3 blockade further enhanced NK cytotoxicity versus Neuro-2a, with enhanced 15–4P NK cell degranulation, Eomesodermin, TRAIL and Fas Ligand expression observed. In vivo, the combination of 15–4P stimulated allogeneic NK cells and TIM-3 blockade after allo-HCT resulted in prolonged survival against NBL with decreased tumor burden compared with NK cells or anti-TIM-3 alone. Depletion of NK cells, but not T cells, abrogated the GVT effect.

Conclusion

Allo-HCT can be a platform for treating NBL using combination ex vivo stimulated allogeneic NK cell therapy with TIM-3 blockade to enhance the GVT effect without inducing GVHD.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• T cell-depleted (TCD) allogeneic hematopoietic cells transplant (allo-HCT) has potential to be a salvage therapy for relapsed/refractory neuroblastoma (NBL) through the graft-versus-tumor (GVT) effect; however, the high incidence of graft-versus-host disease (GVHD) and disease progression has hindered widespread clinical application. Allogeneic natural killer (NK) cell therapy can be a safe and feasible treatment but has had limited efficacy in NBL. T-cell immunoglobulin and mucin-domain containing-3 (TIM-3) blockade has shown encouraging results for a variety of tumors but has not been explored for NBL nor used to enhance the GVT effect.

WHAT THIS STUDY ADDS

• Ex vivo stimulated allogeneic NK cells demonstrate robust expansion, proliferation, activation and cytotoxicity against NBL which is further augmented by the addition of TIM-3 checkpoint blockade. In an allo-HCT model of NBL relapse, mice treated with interleukin-15/IL-15 receptor alpha (IL-15/IL-15Rα) aggressive variant of the Neuro-2a murine neuroblastoma cell line (15–4P) stimulated NK cells and anti-TIM-3 exhibited prolonged overall survival and reduced tumor growth without exacerbating GVHD.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Relapse of NBL after TCD allo-HCT can be successfully treated with the combination of adoptively transferred 15–4P stimulated allogeneic NK cells and TIM-3 checkpoint blockade, representing a novel therapeutic approach that can be translated to the clinic or used as a platform to treat other solid tumors with high TIM-3 ligand expression.

Introduction

High-risk neuroblastoma (NBL) is an aggressive extracranial solid tumor of neural crest origin that commonly occurs in children typically under 5 years old. About 40% of patients with NBL are considered high risk.¹ Unfavorable molecular characteristics indicate poor prognosis, requiring intensive treatment regimens including chemotherapy, surgery, consolidation with autologous hematopoietic cell transplant (auto-HCT), radiation therapy and post-consolidation anti-GD2 immunotherapy with Granulocyte-Macrophage Colony-Stimulating Factor (GM-CSF) and cis-retinoic acid. The addition of anti-GD2 therapy has increased the 5-year overall survival (OS) of high-risk NBL from 20% to 50%, yet half of patients still relapse.² There is still a critical need for the development of therapies for NBL with lower toxicities.³

Auto-HCT has been standard of care for high-risk NBL for decades but offers little benefit outside of stem cell rescue. Three phase III randomized clinical trials have reported that supratherapeutic chemotherapy with auto-HCT support improved event-free survival (EFS) but failed to ameliorate 5-year OS, which highlights a critical need for incorporating novel therapies for the treatment of high-risk NBL.^{4 5} Allogeneic HCT (allo-HCT) offers a unique advantage to auto-HCT due to donor-derived lymphocytes. However, it has been difficult to translate these results to solid tumors due to the high incidence of graft-versus-host disease (GVHD) and disease progression.⁶

Advances in graft engineering now allow for precise removal of allogeneic T cells (eg, αβ-T cell receptor/TCR depletion, CD3 depletion) that can greatly reduce the risk of GVHD; however, these methods increase the likelihood of relapse, infection, and delays in immune reconstitution.^{7 8} Adoptive cell transfer of natural killer (NK) cells after T cell-depleted allo-HCT has the potential to improve outcomes by treating NBL relapse while controlling viral infections without inducing GVHD. NK cells are innate lymphoid cells that play an important role in antitumor surveillance by identifying infected, stressed, or tumorigenic cells through a balance of activating and inhibitory cell surface receptors.^{9 10} NK cells are not limited by antigen recognition; instead, missing or mismatched binding between killer-cell-immunoglobulin and “non-self” major histocompatibility complex (MHC) class I (or human leukocyte antigens class I) triggers a cytotoxic response.¹¹ NK cells do not induce GVHD and can exert a protective defense against GVHD by suppressing allo-reactive T cells.¹² Clinical studies in NBL have confirmed the safety profile of allogeneic NK cell therapy; however, response rates remain modest. Due to the biological limitations of naïve NK cells, methods of ex vivo expansion and activation that enhance NK cell cytotoxicity without inducing exhaustion are critical. Cytokine stimulation with multiple cytokines, including interleukin (IL)-15, has been shown to induce T-cell immunoglobulin and mucin-domain containing-3 (TIM-3) expression on NK cells, suggesting that TIM-3 expression marks a mature and highly activated subpopulation.¹³ TIM-3 expression increases following cytokine stimulation and promotes interferon (IFN)-γ production, and NK cells with high TIM-3 expression are highly responsive.¹⁴ However, cross-linking of TIM-3 with agonist antibodies or cognate ligands significantly inhibits NK cell-mediated cytotoxicity induced by CD16 and NK cell activation and dampens degranulation, making it an ideal target for immune checkpoint inhibition.¹⁵

Here, we explore the use of an aggressive variant of the Neuro-2a murine neuroblastoma cell line (AgN2a-4P), a murine neuroblastoma cell line engineered from an AgN2a to express four immune costimulatory proteins CD54, CD80, CD86, and CD137L (4P) on its surface. Notably, CD137L (41BBL) serves as a well-established NK cell activator, driving proliferation and heightened antibody-dependent cellular cytotoxicity via Fc receptor signaling.¹⁶ Additionally, CD54 can bind to LFA-1 and MAC-1 on NK cells and augments cytotoxicity by promoting adhesion and granule polarization. The current investigation explores the impact of adoptive cell therapy of ex vivo AgN2a-4P-stimulated allogeneic NK cells with TIM-3 immune checkpoint blockade for the first time after allo-HCT for NBL as a novel combination immunotherapy.

Methods

Mice

Female C57BL/6NCr (B6, H-2b), and B6AJF1 (B6AJ, H-2b × H-2a) mice were purchased from the National Cancer Institute Animal Production Program and Charles River Laboratories International (Frederick, Maryland, USA). Mice were housed in accordance with the Guide for Care and Use for Laboratory Mice and used between 8 and 16 weeks of age at time of experimentation.

Murine ex vivo NK isolation and activation

Murine bone marrow cells (BMCs) were harvested from tibias and fibulas and processed into a single-cell suspension. Splenocytes were processed into a single-cell suspension. BMCs and splenocytes were pooled. Red blood cells were lysed with ACK lysis buffer (Lonza, Walkersville, Maryland, USA). NK cells were isolated using the NK cell isolation kit (Miltenyi Biotec, Auburn, California, USA) by negative selection on the AutoMACs Pro (Miltenyi Biotec). NK cells were expanded and activated in the presence or absence of irradiated (100 Gy) AgN2a-4P at a 1:1 ratio and cultured in Roswell Park Memorial Institute (RPMI) medium (Corning Life Sciences, Durham, North Carolina, USA) containing 10 ng/mL IL-15/IL-15 receptor alpha (IL-15Rα) conjugate, comprised of recombinant mouse IL-15 protein (R&D Systems, Minneapolis, Minnesota, USA) and recombinant mouse IL-15Rα Fc chimera protein (R&D Systems), 10% fetal bovine serum (GeminiBio, Sacramento, California, USA) with 50 IU/mL penicillin/streptomycin (Lonza), 0.11 mM 2-beta-mercaptoethanol (Gibco, Carlsbad, California, USA), 1× MEM non-essential amino acids (Corning), 25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer (Corning), and 1 mM sodium pyruvate (Corning). NK cells were expanded for 12 days and were supplemented with additional 10 ng/mL IL-15/IL-15Rα and 5 mL of media every 2 days and maintained at 37°C in 5% CO₂.

Allogeneic bone marrow transplant

BMCs were harvested from B6 mice and processed into single-cell suspensions. T cells were depleted using a CD3 MicroBead Kit (Miltenyi Biotec) by negative selection on the AutoMACs Pro Separator (Miltenyi Biotec) at day+0. Separately, T cells were isolated using a pan T Cell Isolation Kit (Miltenyi Biotec) on single-cell suspensions obtained from splenocytes isolated from B6 mice. B6AJF1 recipient mice received a single dose of 11 Gy total body irradiation using an X-Rad 320 (Precision X-ray Irradiation, Madison, Connecticut, USA). Irradiated recipients received retroorbital injection of 5×10⁶ donor-derived BMCs and 1×10⁶ T cells in serum-free RPMI (Invitrogen, Carlsbad, California, USA).

In vivo tumor challenge and immunotherapy treatment

At day +7, 2×10⁶ Neuro2A NBL cells (H-2a) were prepared as a single cell suspension in serum-free RPMI and injected into B6AJF1 recipient mice subcutaneously on the right hind limb flank to mimic post-HCT relapse. Tumor length and width were measured biweekly using a digital caliper, and tumor growth was calculated as V=½ (length×width²).

Statistical analysis

Statistics were performed using GraphPad Prism V.9.0 (GraphPad Software, San Diego, California, USA). Survival curves were plotted using Kaplan-Meier estimates, and the Mantel-Cox log-rank test was used to analyze survival data. The two-tailed Mann-Whitney U test (non-parametric data sets) or unpaired t-test (parametric data sets) with Welch’s corrections was used for the statistical analysis when comparing two groups. When comparing three or more groups, the Kruskal-Wallis (non-parametric data sets) with Dunn’s multiple comparison post hoc test or one-way analysis of variance (parametric data sets) with Tukey post hoc test. P value<0.05 was considered statistically significant.

Refer to online supplemental methods for more information.

Results

AgN2a-4P stimulated ex vivo activated NK cells show increased expression of activation and proliferation markers

AgN2a-4P is a genetically engineered murine NBL cell line that expresses CD54, CD80, CD86, and CD137L co-stimulatory molecules but has not been tested on allogeneic NK cells previously. AgN2a-4P can interact with NK cells through CD137L, binding CD137, and CD54, binding LFA-1/MAC-1 on NK cells¹⁷ (figure 1A). To study the impact of AgN2a-4P exposure on NK cell expansion and activation, NK cells from B6 mice were expanded with AgN2a-4P cells with IL-15/IL-15Rα at a 1:1 ratio and assessed by flow cytometry at day 12 (figure 1B). B6 NK cells expanded with IL-15/IL-15Rα and AgN2a-4P (15–4P) exhibited a significant increase in the percentage of Natural Killer Group 2, Member D positive (NKG2D⁺) cells (29.27±1.894) compared with IL-15 NK cells (20.80±0.7095). While 15–4P did not impact the percentage of NKp46⁺ NK cells (35.08±1.427) in comparison to IL-15 NK cells (33.70±0.3342), the expression of NKp46 by median fluorescence intensity (MFI) increased (p=0.0286) (figure 1C).

Figure 1. 15–4P Stimulation increases NKG2D expression and murine NK cell proliferation compared with IL-15/IL-15Rα stimulation alone. (A) Schematic showing AgN2a-4P and NK cell receptor-ligand interactions. B6 NK cells are isolated and cultured with IL-15/IL-15Rα conjugate and with or without irradiated AgN2a-4P cells at a 1:1 ratio for 12 days. All expression markers were examined on CD3⁻NK1.1⁺ NK cells. (B) Dot plots of NKG2D (top) and NKp46 (bottom) on IL-15 NK cells (left) and 15–4P stimulated NK cells (right) are shown with percentage of gated cells. (C) Per cent of NK cells positive for NKG2D (left) and NKp46 (right) and MFI below. (D) Fold change of IL-15 NK cells and 15–4P stimulated NK cell expansion at day 12 compared with day 0. (E) Dot plots of proliferation marker Ki-67 on IL-15 NK cells (left) and 15–4P stimulated NK cells (right) are shown with (F) percentage and MFI. (G) Percentage and MFI of TRAIL (top) and Fas-L (bottom) were assessed for each group. Data are representative of an experiment that was replicated two times. Representative dot plots examples of one replicate out of a minimum of three replicates. All bar graphs show individual experimental replicates plotted with mean and SEM (n=3–5). Two-sided two-sample t-tests were performed where indicated (*p<0.05, **p<0.01, ***p<0.001, ns=not significant). AgN2a, aggressive variant of the Neuro-2a murine neuroblastoma cell line; MFI, median fluorescence intensity; NK, natural killer.

 CD80 and CD86 both bind to CD28 and cytotoxic T-lymphocyte associated protein 4 (CTLA-4), which have opposing functions as a stimulatory receptor and inhibitory receptor, respectively. Primary resting human NK cells have been shown to express a CD28 homolog which activates strong lysis against VISTA/B7-H5⁺ targets on synergy with NKp46 and 2B4.¹⁸ Cytokine stimulation induces expression of both CTLA-4 and CD28 in murine NK cells which can activate or inhibit IFN-γ release on ligand binding.¹⁹ To determine the roles of CD80, CD86, CD54, and CD137L in enhancing 15–4P stimulated NK cell responses, NK cells were expanded in the presence of 15–4P in combination with isotype control, anti-CD54 alone, anti-CD137L alone, and anti-CD80/CD86/CD54/CD137L. When cultured with antibodies against all four receptors, there was a significant reduction in the percentage of NKG2D⁺ and NKp46⁺ NK cells compared with isotype control. When cultured with antibody against CD137L, there was no significant reduction in the percentage of NKG2D⁺ and NKp46⁺ NK cells compared with isotype control. When cultured with antibody against CD54, there was a significant reduction in NKp46⁺ NK cells, but not NKG2D⁺ NK cells compared with isotype control. This data supports the importance of NK cell exposure to the co-stimulatory molecule CD54 in the enhanced activation of NK cells (online supplemental figure 1). Due to the expression of 41BBL on AgN2a-4P, we confirmed whether AgN2a-4P exposure during expansion impacts NK cell proliferation. We observed a 78-fold expansion in 15–4P (78.33±20.46) compared with IL-15 NK cells (27.39±1.207) (figure 1D). There was a significant increase (p=0.0006) of Kiel-67 (Ki-67⁺) 15–4P NK cells (52.24±3.953) and Ki-67 expression by MFI compared with IL-15 NK cells (13.42±0.318) (figure 1E, F).

To examine effector capacity, expansion with 15–4P resulted in a significantly higher percentage of TNF-Related Apoptosis-Inducing Ligand (TRAIL⁺) NK cells (19.06±0.6823) compared with IL-15 NK cells (14.10±0.6532) (p=0.0006), and an increase in TRAIL expression by MFI (p=0.0238). Conversely, 15–4P stimulated NK cells exhibited a significant decrease in Fas Ligand (Fas-L⁺) cells (3.998±0.2588) in contrast to IL-15 NK cells (5.935±0.6996), and there was no difference in Fas-L expression by MFI (figure 1G). These findings indicate that exposure to AgN2a-4P during IL-15/IL-15Rα-based NK cell expansion results in comparable activation as NK cells are likely to be maximally activated through expansion with IL-15. Increased proliferation markers were observed in 15–4P stimulated NK cells compared with NK cells expanded with IL-15/IL-15Rα alone. Additionally, the potential impact of cytotoxicity depending on if TRAIL versus Fas-L pathways are used to trigger apoptosis of NBL tumors requires further investigation.

The NBL tumor microenvironment (TME) is highly immunosuppressive, influencing NK cells through a complex interplay of activating and inhibitory receptor signaling. TIM-3, a marker of maturation in NK cells, interacts with its cognate ligands, including Carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM-1), Galectin-9 (Gal-9), High Mobility Group Box 1 (HMGB1), and Phosphatidylserine (PtdSer) to exert inhibitory regulation in NK cells. We verified the surface expression of MHC I and galectin-9 and CEACAM1 on murine NBL cell lines Neuro2a and NXS2. Neuro2a displayed a significantly higher percentage of CEACAM1⁺ cells (20.08%±0.9016) (p=0.002), and Gal-9⁺ cells (95.10%±0.4235) (p=0.002) compared with NXS2 (figure 2A). Both Neuro2a and NXS2 had low expression of MHC-I. The MFI of CEACAM1 (p=0.0047) and galectin-9 (p=0.0049) was also notably higher on Neuro2a cells compared with NXS2 (figure 2A). Due to their intracellular expression, HMGB-1 and PtdSer messenger RNA (mRNA) were measured. Neuro2a showed significantly higher mRNA expression of both HMGB-1 and PtdSer compared with NXS2 (online supplemental figure 2A). NXS2 lost expression of CEACAM1, Gal-9, and HMGB-1 when implanted in a mouse flank (online supplemental figure 2B). Neuro2A retained expression of CEACAM1, Gal-9, and HMGB-1 when implanted in a mouse flank (online supplemental figure 2C).

Figure 2. 15–4P stimulated B6 NK cells show limited efficacy against neuroblastoma in vitro and increased expression of TIM-3. (A) NBL cell lines, NXS2 and Neuro2A, were interrogated for the presence of tumor ligands CEACAM-1, galectin-9, MHC-I and with percentage (top) and MFI (bottom) shown for each marker. (B) IL-15 NK cells (black) or 15–4P stimulated NK cells (blue) were plated with neuroblastoma Neuro2A cells at an effector-to-target ratio of 10:1 with green fluorescent caspase 3/7-substrate for 24 hours. Total green object area (μm²/well) was measured by IncuCyte real-time analysis and AUC is shown for all groups. Mean with SEM shown for experimental replicates (n=3 per group). (C) Fas (top) and TRAIL-R2 (bottom) per cent positive cells and MFI are shown for NX2S and Neuro2a. (D) Representative dot plots for subsets of TIM-3⁺ IL-15 NK cells (left) and 15–4P stimulated NK cells (right) are shown. (E) Percentage (top), and MFI (bottom) are shown for the presence of TIM-3⁺ on IL-15 NK cells (left) and 15–4P stimulated NK cells (right). All the data are representative of an experiment that was replicated two times. Mean with SEM shown for experimental replicates (n=6 per group). Two-sided two-sample t-tests were performed (*p<0.05, **p<0.01, ns=not significant). AgN2a, aggressive variant of the Neuro-2a murine neuroblastoma cell line; AUC, area under the curve; MFI, median fluorescence intensity; MHC, major histocompatibility complex; NBL, neuroblastoma; NK, natural killer.

Next, we evaluated the efficacy of 15–4P-stimulated ex vivo activated allogeneic B6 NK cells (H-2b) against Neuro2a (H-2a) murine NBL to evaluate the impact of Ly49 mismatch on cytotoxicity. Neuro2a NBL targets were co-cultured with either IL-15/IL-15Rα expanded allogeneic NK cells or 15–4P expanded allogeneic NK cells for 24 hours. Apoptotic Neuro2a cells were selected by size and fluorescence intensity of caspase 3/7 and quantified using live-cell imaging by the IncuCyte system. 15–4P-expanded allogeneic NK cells exhibited slightly increased target killing compared with IL-15-expanded allogeneic NK cells, despite showing a significant decrease in Fas-L expression (figure 2B).

Surface TIM-3 upregulation was confirmed on IL-15/IL-15Rα NK cells and 15–4P stimulated NK cells. Results showed a significantly higher percentage of TIM-3⁺ cells (20.78%±1.695) (p=0.0006) and TIM-3 expression by MFI in 15–4P stimulated NK cells relative to IL-15/IL-15Rα NK cells (4.134%±0.1592) (p=0.0012) (figure 2D, E). The elevated expression of TIM-3 on AgN2a-4P NK cells combined with elevated expression of putative TIM-3 ligands on NBL cells suggests this immune checkpoint may be a relevant axis governing NK cell function.

15-4P stimulated NK cells show enhanced cytotoxicity against NBL in vitro after TIM-3 immune checkpoint blockade

Initially, we measured IFN-γ and tumor necrosis factor (TNF)-α mRNA levels to determine the impact of TIM-3 blockade on NK cell cytokine production after ex vivo expansion. We co-cultured IL-15 B6 allogeneic NK cells or 15–4P stimulated B6 allogeneic NK cells with isotype control or anti-TIM-3 antibody in the presence of Neuro2a NBL cells at an effector-to-target (E:T) ratio of 10:1 and assessed cytokine production. TNF-α production was significantly increased in the 15–4P NK cells compared with IL-15 NK cells; however, there was no increase after 15–4P NK cell with TIM-3 blockade after exposure to Neuro2a NBL. Similarly, we observed a notable increase in IFN-γ production in 15–4P stimulated NK cells compared with IL-15 NK cells, but 15–4P NK cells showed no further increase in IFN-γ (online supplemental figure 2D). Then, to investigate whether blocking TIM-3 could enhance the killing potential of 15–4P stimulated NK cells, we co-cultured 15–4P stimulated NK cells with Neuro2a at an E:T ratio of 10:1 with isotype control or anti-TIM-3 and then assessed surface expression of activating receptors and death ligands as well as cytotoxicity. There was a high percentage of NKG2D⁺ and NKp46⁺ cells and elevated surface expression of NKG2D and NKp46 by MFI among 15–4P stimulated NK cells as seen in prior experiments, but these subsets were not enriched by TIM-3 blockade, suggesting NK cells were already maximally activated. After exposure to Neuro2a, 15–4P NK cells with TIM-3 blockade exhibited a higher percentage (p=0.004) but not MFI (p=0.05) of TRAIL than without the anti-TIM-3 antibody. Compared with 15–4 P cells alone, TIM-3 blockade enhanced both the percentage (p=0.02) of Fas-L⁺ 15–4P NK cells, and Fas-L MFI (p=0.007) compared with 15–4P NK cells after Neuro2a exposure (figure 3A). This experiment was repeated with NXS2 NBL cells. Interestingly, 15–4P NK cells treated with TIM-3 blockade showed no difference in NKG2D and NKp46 expression by MFI as compared with 15–4P NK cells (online supplemental figure 3A). Additionally, in the presence of NXS2, we found no significant increase in TRAIL, Fas-L, granzyme B or perforin per cent or expression, suggesting that blockade of exposure to TIM-3 ligands on NBL is required to augment allogeneic NK cell activation (online supplemental figure 3B).

Figure 3. TIM-3 blockade enhances 15–4P stimulated NK cells cytotoxicity against NBL cells expressing TIM-3 ligands. 15–4P stimulated NK cells with isotype control (blue) or anti-TIM-3 blockade (red) were plated with Neuro2a NBL at an E:T ratio of 10:1 for 4 hours. NK cells were collected and analyzed by FACs. (A) NKG2D, NKp46, TRAIL, and Fas-L percentage positive (top) and MFI (top) for NK cells are shown. (B) NK cell degranulation was measured by CD107a percentage positive (top) and MFI (bottom) for 15–4P stimulated NK cells (blue) or 15–4P stimulated NK cells with anti-TIM-3 blockade (red). (C) Plots comparing percentage and MFI of granzyme B and perforin for 15–4P stimulated NK cells in the presence or absence of anti-TIM-3 blockade or IgG2a isotype control are shown. (D) Percentage and MFI of Eomes and T-bet are shown for all groups as well as (E) ratio of the per cent of Eomes⁺ NK cells relative to the per cent of T-bet⁺ NK cells. Mean with SEM shown for experimental replicates (n=6). (F) For cytotoxicity experiments, IL-15 NK cells with isotype control (black) or anti-TIM-3 blockade (green), and 15–4P stimulated NK cells with isotype control (blue) or TIM-3 blockade (red) were plated with Neuro2a NBL cells at an E:T ratio of 10:1 with green fluorescent caspase 3/7-substrate for 24 hours. (G) Cytotoxicity of 15–4P stimulated NK cells with isotype control or TIM-3 blockade against NXS2 NBL is shown. Cytotoxicity was determined by caspase 3/7 (total green object area (μm²/well)) tracking of tumor cell death with IncuCyte real-time analysis, and AUC was calculated for all groups. Two-sided two-sample t-tests were performed for comparison between two groups. Comparisons between three or more groups were analyzed with Brown-Forsythe and Welch one-way ANOVA test with Dunnett’s T3 multiple comparisons test (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, ns=not significant). AgN2a, aggressive variant of the Neuro-2a murine neuroblastoma cell line; ANOVA, analysis of variance; AUC, area under the curve; Eomes, Eomesodermin; E:T, effector-to-target; MFI, median fluorescence intensity; NBL, neuroblastoma; NK, natural killer.

Next, we evaluated cytotoxicity by quantifying CD107a degranulation on allogeneic NK cells following exposure to NBL. Interestingly, we found no discernible difference in degranulation between 15–4P stimulated NK cells with or without TIM-3 blockade in response to NXS2, which have lower expression of TIM-3 ligands (online supplemental figure 3C). However, TIM-3 blockade resulted in a significant increase in the percentage of CD107a+15–4P NK cells, as well as MFI of CD107a granules (figure 3B), in response to Neuro-2a. Additionally, we observed a significant increase in the percentage of granzyme B-positive cells following TIM-3 blockade, as well as a significant increase in both granzyme B and perforin expression (figure 3C).

To determine the impact of TIM-3 blockade on NK cell maturation, we assessed the expression of the transcription factors T-box expressed in T cells (T-bet) and Eomesodermin (Eomes) after exposure to NBL tumor cells. TIM-3 blockade did not impact T-bet per cent or expression as 15–4P NK, and 15–4P NK with TIM-3 blockade, cells showed similarly low levels of T-bet. However, we observed a significant rise in both the percentage (p=0.004) and MFI (p=0.004) of Eomes in 15–4P stimulated NK cells after exposure to TIM-3 blockade (47.05±1.607) compared with 15–4P NK cells alone (38.33±0.6756) (p=0.0018). This was consistent with Eomes MFI (figure 3D). We determined the ratio of Eomes positive cells relative to T-bet in 15–4P stimulated NK cells and 15–4P stimulated NK cells with anti-TIM-3 after Neuro2a NBL exposure; however, there was no significant difference between the two groups (figure 3E). Interestingly, TIM-3 blockade did not enhance Eomes expression in 15–4P NK cells after exposure to NXS2 NBL cells which have lower expression of TIM-3 ligands (online supplemental figure 3D). These findings suggest that TIM-3 blockade enriches for NK cells in early maturation.²⁰

Following treatment with anti-TIM-3, 15–4P stimulated NK cells exhibited heightened expression of death ligands and increased markers of degranulation, indicating a phenotype conducive to antitumor activity. Allogeneic NK cells were incubated with Neuro2a NBL with anti-TIM-3 or an isotype control, and apoptotic NBL cells were quantified using IncuCyte live cell imaging of caspase 3/7 expression. For all groups, peak killing occurred after the 8-hour mark. 15–4P NK cells treated with anti-TIM-3 antibody exhibited significantly higher killing compared with all other groups (figure 3F). The enhanced killing capacity of 15–4P NK cells was also observed in the presence of a second murine NBL cell line, 9464D (online supplemental figure 3E). This effect was not observed against NXS2 (figure 3G).

TIM-3 inhibition modulates gene expression in NK cells

We conducted a comparative evaluation of IL-15 allogeneic NK cells and 15–4P expanded allogeneic NK cells following exposure to the NBL TME to investigate whether TIM-3 blockade modulated NK cell gene expression. Total RNA was isolated from 15 to 4P stimulated NK cells NK cells co-cultured with Neuro2a in the presence of isotype controls or anti-TIM-3 for 4 hours. Using the NanoString/nCounter Mouse PanCancer Immune Profiling Panel, the expression of 760 genes was examined. The comparison between 15–4P stimulated NK cells and 15–4P stimulated NK cells treated with TIM-3 blockade showed 25 significantly upregulated genes, and 12 genes showed significant downregulation compared with 15–4P stimulated NK cells treated with an isotype control. Within the upregulated subset of genes, numerous genes associated with NK cell trafficking and target cell recognition (CCL4, CCL3, CCL1, XCL1, CXCR6), and the non-canonical NF-κB pathway (LTA, LTB, TNFSF10, TNFSF14, TNFSF8, RELB, TNF, TRAF3)²¹ were observed. Both ICAM1, which encodes CD54, and TNFSF10, which encodes TRAIL, exhibited a fold change of 2.4 (with false discovery rate p values of 0.003 and 0.0024, respectively). Furthermore, genes encoding IL-15 and IL-21 receptors were upregulated. Surprisingly, Prf1 (perforin) was downregulated in IL-15/AgN2a-4P/TIM-3 NK compared with 15–4P stimulated NK cells despite TRAIL upregulation (figure 4A, B; online supplemental table 1). This alteration may suggest functional impairment due to exhaustion. Following the initial granule-mediated killing phase, NK cells may switch to death receptor signaling to sustain serial killing on granule depletion.^{22 23}

Figure 4. 15–4P stimulated NK cells treated with anti-TIM-3 blockade respond against NBL through multiple cytotoxicity pathways. (A) Volcano plots representing the −log10 FDR-adjusted p values and log2 fold changes for differentially expressed genes within the comparison of 15–4P NK cells with isotype control versus 15–4P NK cells with anti-TIM-3 antibody (right) (n=2). The significance threshold for an FDR of ≤0.05 is represented by the horizontal line. Vertical lines represent genes with absolute fold change of ≥2. Upregulated genes are shown as green dots and downregulated genes are shown as red dots. (B) Upregulated DEGs from the comparison of 15–4P NK cells with isotype control versus 15–4P NK cells with anti-TIM-3 antibody, exhibiting an absolute fold change of ≥2, are plotted with respective fold change and p values. (C) 15–4P stimulated NK cells with anti-TIM-3 blockade (green) and anti-NKG2D blockade, anti-Fas-L blockade, or anti-TRAIL blockade were plated with Neuro2a NBL cells at an E:T ratio of 10:1 with green fluorescent caspase 3/7-substrate for 24 hours. Cytotoxicity and AUC (D) are shown for all groups. All the data are representative of an experiment that was replicated two to three times. Mean with SEM shown for technical replicates (n=5 per group). Comparisons between three or more groups were analyzed with Brown-Forsythe and Welch one-way ANOVA test with Dunnett’s T3 multiple comparisons test (****p<0.0001). AgN2a, aggressive variant of the Neuro-2a murine neuroblastoma cell line; ANOVA, analysis of variance; AUC, area under the curve; DEG, differentially expressed gene; E:T, effector-to-target; FDR, false discovery rate; NBL, neuroblastoma; NK, natural killer.

NK cells use multiple cytotoxic pathways in tumor surveillance, transitioning between granule release and death receptor-mediated killing to initiate and sustain long-term cytotoxic activity. To assess the impact of multiple NK cell cytotoxicity pathways on the killing efficiency of 15–4P stimulated NK cells in the presence of TIM-3 blockade, we exposed allogeneic NK cells to Neuro2a NBL in the presence of blocking antibodies targeting NKG2D, TRAIL, and Fas-L. The inhibition of all three targets resulted in a significant reduction in NK cell killing. Notably, the blockade of Fas-L completely ameliorated NK cell cytotoxicity against NBL compared with NKG2D or TRAIL blockade, suggesting there is a hierarchy of pathways used to mediate NK cytotoxicity of NBL when TIM-3 is engaged (figure 4C). These data underscore the potential of TIM-3 blockade to enhance NK cell cytotoxicity against NBL, particularly in tumors with high expression of TIM-3 ligands, emphasizing the importance of considering the makeup of the NBL TME in therapeutic strategies.

Combination therapy of 15-4P stimulated allogeneic NK cells with anti-TIM-3 reduces NBL progression after allo-HCT

Given our observations regarding the positive impact of TIM-3 blockade on the phenotype and function of 15–4P stimulated NK cells in vitro, we hypothesized that this therapy could yield therapeutic benefits in vivo. However, the intricate pro-inflammatory environment post-allo-HCT and the immunosuppressive NBL TME present additional challenges that could impede NK cell cytotoxic efficacy. To address this concern, we evaluated the potency of the combination allogeneic NK cell treatment against Neuro2a NBL following allo-HCT. Initially, we established an MHC-mismatched haploidentical allo-HCT model wherein lethally irradiated B6AJF1 (H-2b × H-2a) recipients received 5×10⁶ T cell-depleted (TCD) B6 (H-2b) bone marrow with 1×10⁴ B6 T-cell add back on day+0 to simulate residual T cells present in the donor graft during clinical T-cell depletion protocols (eg, CD3 or αβ-TCR depletion). On day+7, mice were subcutaneously inoculated with Neuro2a (H-2a) cells to mimic relapse post-HCT. Recipients then were infused with 3 weekly infusions of B6 IL-15 allogeneic NK cells or B6 15–4P stimulated allogeneic NK cells administered with either an isotype control or anti-TIM-3 treatment every 7 days starting on day+9 (figure 5A). Although TIM-3 blockade has not been tested after haploidentical allo-HCT, despite the presence of residual T cells in the donor bone marrow graft, we did not observe significant differences in GVHD-associated weight loss (figure 5B). By day 35, mice treated with 15–4P NK cells and anti-TIM-3 exhibited the most substantial reduction in tumor growth compared with controls (figure 5C). Treatment with 15–4P allogeneic NK cells combined with TIM-3 blockade significantly prolonged OS compared with mice treated with 15–4P stimulated NK cells and an isotype control, anti-TIM-3 alone, IL-15 NK cells alone, IL-15 NK cells alone with anti-TIM-3, or no treatment groups (figure 5D).

Figure 5. TIM-3 blockade with 15–4P stimulated allogeneic NK cells to increase GVT without inducing GVHD after allo-HCT. (A) Schematic shown for post allo-HCT treatment of NBL relapse. (B) Per cent weight change for mice treated with alloBMT and IL-15 B6 NK cells or 15–4P stimulated B6 NK cells with isotype or TIM-3 blockade is shown (n=5 mice per group). (C) Overall survival for mice receiving alloBMT and no further treatment (NT) or treatment with three infusions of IL-15 B6 NK cells or 15–4P stimulated B6 NK cells with isotype or TIM-3 blockade is shown (n=5 mice per group). (D) Tumor volume shown for with alloBMT and IL-15 B6 NK cells or 15–4P stimulated B6 NK cells with isotype or TIM-3 blockade is shown (n=5 mice per group). All the data are representative of an experiment that was replicated two times. Tumor volume was calculated as V=½(length×width²). Bars and points with error bars show mean with SEM. One-way ANOVA and Kruskal-Wallis with Dunn’s multiple comparisons tests were performed (*p<0.05, **p<0.01, ***p<0.001, ns=not significant). AgN2a, aggressive variant of the Neuro-2a murine neuroblastoma cell line; allo-BMT, allogeneic bone marrow transplant; allo-HCT, allogeneic hematopoietic cell transplant; ANOVA, analysis of variance; BMC, bone marrow cell; GVHD, graft-versus-host disease; GVT, graft-versus-tumor; NBL, neuroblastoma; NK, natural killer.

Given the known role of TIM-3 blockade in promoting T-cell responses and the presence of a low dose of T cells in the graft, we sought to delineate the contributions of NK cells and T cells to graft-versus-tumor (GVT) effects and OS in our allo-HCT model by depletion of these cell subsets prior to treatment. In our T-cell depletion studies, B6AJF1 mice received T-depleted B6 bone marrow without T-cell add-back. In NK depletion studies, B6AJF1 mice received NK-depleted B6 bone marrow with T-cell add-back. Since NK cells typically are the first lymphocyte to reconstitute after allo-HCT, anti-NK1.1 antibody was administered intraperitoneally every 4 days to maintain NK cell depletion. NK1.1 depletion was performed to ensure the successful engraftment of allogeneic NK cells as endogenous NK cells can compete for space within the bone marrow niche and induce rejection of donor NK cells due to alloreactivity.^{24 25} Subsequently, mice were inoculated with Neuro2A cells and treated with anti-TIM-3 antibody alone or B6 15–4P stimulated allogeneic NK cells with anti-TIM-3 antibody (figure 6A). Consistently, there were no noticeable differences concerning GVHD-associated weight loss (figure 6B). NK cell depletion led to a significant decrease in OS, while TCD mice exhibited significantly increased survival like that of mice treated with 15–4P stimulated NK cells and TIM-3 blockade (figure 6C). Furthermore, NK cell depletion resulted in larger tumors in all mice, with no significant difference compared with the control, whereas T-cell depletion led to a significant decrease in tumor growth (figure 6D). These findings suggest that while T cells are present in the graft, NK cells may play a more substantial role in the survival benefit conferred by 15–4P stimulated NK cells, possibly due to the limited number of T cells compared with NK cells resulting from repeated adoptive transfers. Moreover, these results underscore the crucial role of NK cells as the primary drivers of TIM-3-induced therapeutic benefits against NBL after allo-HCT, given that similar outcomes were not observed with anti-TIM-3 treatment alone (online supplemental figure 4A, B).

Figure 6. NK cells are necessary for the GVT effect mediated by TIM-3 blockade against NBL relapse after allo-HCT. (A) Schematic shown for post alloBMT treatment of NBL relapse with NK and T-cell depletion. (B) Per cent weight change, (C) overall survival and (D) tumor volume is shown for mice receiving alloBMT with no further treatment or treatment with three infusions of 15–4P stimulated B6 NK and TIM-3 blockade, T-cell depleted alloBMT with three infusions of 15–4P stimulated B6 NK and TIM-3 blockade, or NK cell depleted alloBMT with TIM-3 blockade. (n=5 mice per group). Tumor volume was calculated by as V=½(length×width²). Bars and points with error bars show mean with SEM. One-way ANOVA and Kruskal-Wallis with Dunn’s multiple comparisons tests were performed (**p<0.01, ***p<0.001, ns=not significant). AgN2a, aggressive variant of the Neuro-2a murine neuroblastoma cell line; allo-BMT, allogeneic bone marrow transplant; allo-HCT, allogeneic hematopoietic cell transplant; ANOVA, analysis of variance; BMC, bone marrow cell; GVT, graft-versus-tumor; NBL, neuroblastoma; NK, natural killer.

Discussion

T cell-depleted (TCD) allo-HCT has recently gained traction as a safe alternative to standard unmanipulated allo-HCT in order to leverage the GVT effect mediated by donor-derived innate effector cells, like NK cells, while mitigating lethal GVHD.²⁶ While graft manipulation techniques like CD3 or αβ-TCR depletion have achieved up to 5 log10 depletion of T cells, there are always depletion-resistant CD8+ or CD4+ T cells to consider toward impacting GVT or GVHD.²⁷ A major limitation of TCD allo-HCT is the increased risk of infectious complications and relapse due to delayed immune reconstitution of T cells. Preclinically, we have previously explored the use of TCD haploidentical allo-HCT in combination with adoptively transferred ex vivo activated NK cells and an anti-GD2 immunocytokine.²⁰ In this study, we found that 15–4P ex vivo expanded NK cells have increased potency against NBL target cells, especially when combined with TIM-3 blockade. CD54 and CD137L co-stimulation have never been combined to activate murine NK cells ex vivo. While CD54⁺CD137L⁺AgN2 a-4P cells were initially designed as an in vivo vaccine for stimulating T cells, we demonstrate for the first time that combining IL-15/IL-15Ra with AgN2a-4P exposure ex vivo significantly boosts NK cell proliferation and activation, allowing for adoptive transfer of highly potent immune effector cells in the allo-HCT setting.

AgN2a-4P stimulation increased both NKG2D and TRAIL expression on NK cells. 15–4P-stimulated NK cells exhibited increased IFN-γ production compared with IL-15/IL-15RA stimulation alone. Prior studies done in a syngeneic model by Jing et al demonstrated that AgN2a-4P vaccination enacted an anti-NBL response through CD8⁺ and CD4⁺ T cells mediated response, despite their tumor target being the parental line AgN2a, which does not express MHC class II.²⁸ While studies have shown that CD4⁺ T cells can kill tumor targets through an indirect mechanism involving IFN-γ directed elimination, the contribution of NK cells to the anti-NBL response cannot be ruled out.²⁹ NK cell reconstitution occurs rapidly following HCT, and it is possible that AgN2a-4P vaccination induced an NK cell-mediated antitumor response and increased IFN-γ production, which indirectly facilitated CD8⁺ T cell and macrophage antitumor responses.²⁸ Additionally, we found that 15–4P stimulation significantly enhanced NK cell-mediated lysis of Neuro2a NBL tumor compared with IL-15 NK cells in vitro. IL-15 exposure results in an upregulation of TIM-3 expression, where TIM-3 can act as a negative regulator of activation to limit autoreactivity in innate immune cells and maintain self-tolerance.^{30 31} Although increased TIM-3 upregulation is associated with activated NK cells, cytotoxicity is significantly impaired on crosslinking of TIM-3 with cognate ligands.¹⁵ In addition to TIM-3 regulation of NK cell activation, inhibitory signals in the NBL TME can dampen the NK cell response through mechanisms including blockade of target cell adhesion and inhibition of LFA-1 activation, subjugating a weaker activating stimulus and tilting the NK cell response towards inactivation.^{32 33} TIM-3 ligands such as Gal-9, CEACAM-1, HMGB1, and PtdSer are expressed at different levels on and within NBL cells and are present in the TME as they are ancillary to pathogenesis.34,36 Our results showed that multiple TIM-3 ligands were expressed at a higher proportion on Neuro2a, but not NXS2, suggesting that TIM-3 ligand exposure is cell line dependent and may only be upregulated in a subset of patients with NBL. However, exposure to inhibitory signals from TIM-3 ligands can originate from tumor-associated macrophages as well as other immune cell types within the TME.³⁷

To better understand the mechanisms driving 15–4P NK cells treated with TIM-3 blockade against NBL, we analyzed upregulation of 25 genes and downregulation of 12 genes in 15–4P stimulated NK cells compared with 15–4P stimulated NK cells alone, including ICAM1 and TNFSF10 encoding CD54 and TRAIL, respectively. Within the upregulated subset of genes, numerous genes were found to be associated with NK cell trafficking and target cell recognition (CCL4, CCL3, CCL1, XCL1, CXCR6) in 15–4P stimulated NK cells with TIM-3 blockade compared to 15–4P NK cells alone. Modulating chemokine receptors may contribute to increased NK cell motility, thereby allowing 15–4P NK cells enhanced ability to track Neuro2a NBL cells in the absence of TIM-3 inhibition.³⁸ Interestingly, we saw increased expression of multiple genes involved in the TNFR2 non-canonical NF-kB pathway (LTA, LTB, TNFSF10, TNFSF14, TNFSF8, RELB, TNF, TRAF3) and in DAP12 signaling (H2-Aa, KLRC2), which suggests the involvement of NKG2D activation in anti-TIM-3 15–4P NK cell mediated killing. Functional cytotoxicity studies confirmed the ability of 15–4P NK cells treated with TIM-3 blockade to kill Neuro2A targets, which were reversed when blocking NKG2D, TRAIL, or Fas-L. Interestingly, we observed a downregulation of Prf1 (perforin) in 15–4P NK cells treated with TIM-3 blockade compared with 15–4P stimulated NK cells, while TNFSF10 (TRAIL) showed upregulation. This may suggest functional impairment in 15–4P NK cells treated with TIM-3 blockade, potentially due to exhaustion. The upregulation of TRAIL and the non-canonical Nuclear Factor kappa-light-chain-enhancer of activated B cells (NF-κB) pathway intermediates suggests a compensatory mechanism, where NK cells engage the death receptor pathway for killing on granule depletion, thereby sustaining serial killing of target cells.²² However, NK cells are regulated by a complex interplay of activating and inhibitory co-receptors which contribute to a dynamic response. Therefore, it cannot be ruled out that unidentified alternative signaling pathways may influence TIM-3 induced NK cell activation. Further investigation is required to understand how TIM-3 engagement mobilizes FAS-L and TRAIL in the context of NK cell biology.

Several studies have indicated that blocking TIM-3 interactions during hepatitis B virus infection, as well as in melanoma, acute myelogenous leukemia and multiple myeloma, resulted in enhanced NK cell cytotoxicity.39,42 Conversely, other studies have reported enhanced IFN-γ production without a corresponding improvement in NK cell degranulation, suggesting that TIM-3 engagement can shift the NK cell response between positive and negative signaling events through differential tyrosine phosphorylation of the cytoplasmic tail.^{14 43 44} We observed increased IFN-γ production in 15–4P stimulated NK cells, which indicated that exposure to AgN2a-4P during ex vivo expansion shifts NK cells to a more immature phenotype that is still capable of robust cytotoxicity. However, the addition of TIM-3 blockade did not further enhance IFN-γ and TNF-α production in 15–4P NK cells. TIM-3 receptor-ligand engagement has been proposed to inhibit the Extracellular signal-regulated kinase (ERK) and NF-κB pathways in NK cells, yet Nuclear Factor of Activated T-cells (NFAT) signaling remains unaffected, thereby enabling NK cells to sustain cytokine production.⁴⁵ Nevertheless, further investigation is warranted to fully elucidate the underlying mechanism behind the effect of TIM-3 on cytokine production in NK cells.

We validated our observations in a haploidentical allo-HCT murine model of NBL relapse to establish preclinically the combination of adoptive transfer of NK cells and TIM-3 blockade treatment. We observed that the combination of NK cell donor lymphocyte infusions and TIM-3 blockade treatment is a viable therapeutic platform. Mice treated with multiple infusions of 15–4P stimulated NK cells and anti-TIM-3 antibody showed significantly smaller NBL tumors and prolonged survival. However, tumors began to grow shortly after treatment ceased. Prior studies show that following IL-15 expansion, allogeneic ex vivo expanded NK cells remain detectable for up to 16 days after infusion in a murine model.⁴⁶ The limited persistence of adoptive NK cell therapies may require frequent infusions to effectively impact clinical outcomes, highlighting a limitation of this therapeutic approach. However, clinical studies have demonstrated favorable outcomes with the use of NK cell infusions combined with immunotherapy as an effective bridge therapy before consolidation.¹⁸ Future studies may aim to generate a sustained NK cell response by designing an in vivo tumor vaccine with costimulatory molecules and/or cytokines specific to activating NK cells.

TIM-3 also plays a crucial role in modulating immune responses during acute GVHD following allo-HCT. Donor T cells in the allogeneic setting rapidly upregulate TIM-3 expression. Inhibition of the TIM-3/Gal-9 interaction has been shown to lead to increased proliferation of T cells and heightened GVHD lethality. However, TIM-3 inhibition after regulatory T-cell depletion has been shown to increase IFN-γ production and reduce GVHD severity.¹⁹ In our experimental model, mice received TCD BMCs with 1×10³ T cells added back to the graft. Notably, GVHD development was absent despite the use of TIM-3 blockade in the presence of residual donor T cells in the bone marrow graft. We then investigated whether the anti-NBL response was driven by NK cells or T cells in the graft. Only NK cell depletion reversed the clinical benefit of TIM-3 blockade, leading to larger tumors and worse survival compared with T-cell depletion, which had no effect. It is possible that the 15–4P NK cell infusions increased IFN-γ production while mediating GVT responses, which indirectly facilitated CD8⁺ T cell and macrophage antitumor responses as well.²⁸ However, it is important to acknowledge the divergence between human and mouse TIM-3 function. In mice, TIM-3 engagement inhibits NK cell activation and cytokine production, primarily by promoting apoptosis of activated NK cells.⁴⁷ In contrast, human NK cells exhibit a gradient response where TIM-3 can elicit co-stimulatory or inhibitory signals based on cellular conditions.⁴⁸ Nevertheless, these results support the conclusion that the anti-NBL response is primarily driven by NK cells, although the potential contribution of blocking TIM-3⁺ macrophages and dendritic cells on NBL clearance needs to be explored.

In summary, this study explored the combination of ex vivo activated NK cells with TCD allo-HCT and revealed the enhanced NK cell potency against relapsed NBL, particularly when paired with TIM-3 blockade. TCD allo-HCT has emerged as a safe and effective treatment for multiple hematologic malignancies, and in combination with cell-based immunotherapy may become a platform for treating solid tumors to maximize the GVT effect without exacerbating lethal GVHD.

Data availability statement

Data are available upon reasonable request.