Introduction

Glioblastoma multiforme (GBM) is the most prevalent and lethal form of primary brain cancer in adults, with over 14,000 new cases diagnosed annually and accounting for 10,000 deaths each year in the USA, resulting in a 5-year survival rate of less than 5%.1 2 The median survival for GBM is 15 months, even with multimodal treatments comprising surgery, radiotherapy, and temozolomide (TMZ). Currently, there are no licensed immunotherapies available for patients with GBM, indicating a significant clinical unmet need.

Recent clinical trials have demonstrated the potential of immunotherapies that target the epidermal growth factor receptor (EGFR) variant III (EGFRvIII), a key GBM-specific antigen, for GBM, marking a significant advance in treating what was once considered an immunologically unresponsive cancer.3–6 Despite these advances, GBM’s antigenic diversity poses a formidable barrier to the effectiveness of such treatments. Trials involving chimeric antigen receptor T cells and vaccines aimed at EGFRvIII have shown limited success, with instances of the tumor evading treatment.3 5 A study highlighting the complexity of GBM identified intratumoral heterogeneity in the expression of EGFR, IL-13Rα2, and HER2 in patient with GBM samples, underscoring the challenge of addressing GBM’s heterogeneity through therapies targeting a single tumor-associated antigen (TAA).7 8 These findings indicate that a single-antigen approach may be insufficient for managing the complex tumor mass of GBM and suggest that strategies targeting multiple antigens could lead to better therapeutic outcomes.

Bispecific T-cell engagers (BTEs) are an emerging tool in cancer immunotherapy, designed to direct the body’s T cells to target and destroy disease-causing cells by simultaneously binding to both TAA and CD3. This approach shows promising potential in treating several types of cancer, marking a significant step forward in leveraging the immune system for oncological therapies.9 In our previous studies, we described two DNA-encoded BTEs (DBTEs) targeting either EGFRvIII or IL-13Rα2, demonstrating tumor suppression in animal studies.10 11 However, their effectiveness was limited, especially in heterogeneous tumor environments. To overcome the challenges posed by the heterogeneity of GBM, we present a novel approach: DNA-encoded tri-specific T-cell engagers (DTriTEs) that simultaneously target both IL-13Rα2 and EGFRvIII. We propose three DTriTE designs, each featuring a unique configuration of three antigen-binding domains (IL-13Rα2, EGFRvIII, and CD3) in a single molecule.

In this manuscript, we show that DTriTEs demonstrated significant efficacy in vitro binding specifically to target cells and activating T cells to release antitumor cytokines such as interferon (IFN)-γ, tumor necrosis factor (TNF)-α, and interleukin (IL)-2. Notably, our constructs also activate natural killer T (NKT) cells, which play a pivotal role in bridging innate and adaptive immunity,12 enhancing the overall immune response against the targeted tumors.

In vivo studies in immunocompromised mouse models reveal that our lead construct, DT2035, achieves sustained expression and effective tumor control in a heterogeneous GBM model. This model, which exhibits a heterogeneous expression of EGFRvIII and IL-13Rα2, effectively resembles the complex tumor environment seen in human GBM, suggesting that DT2035 could offer potential clinical benefits. Moreover, DT2035 has been shown to induce robust T cell-mediated cytotoxicity and immune activation in patient peripheral blood mononuclear cells (PBMCs), demonstrating its potential to reactivate the immune systems of patients who have undergone extensive prior treatments.

These findings highlight the potential of DT2035 as an impactful new approach for treating GBM. By targeting multiple antigens simultaneously, DT2035 suppressed heterogeneous tumors, overcoming the limitations of monovalent therapies. Further research is warranted to explore its full therapeutic potential and to investigate its application to other cancers with similar immunological landscapes.

ResultsDesign and in vitro expression of DTriTEs targeting IL-13Rα2 and EGFRvIII

We developed IL-13Rα2 and EGFRvIII-targeting DTriTE constructs by fusing humanized single-chain variable fragments (scFv) specific for IL-13Rα2, EGFRvIII, and CD3, connected via flexible GS linkers to create a unified chain (figure 1A). To enhance expression, we optimized both codons and messenger RNA (mRNA) sequences of these constructs and incorporated a human IgE leader sequence at the N-terminus, as detailed in our prior work.10 11 13 These constructs were integrated into the pVAX1 vector, a modified expression system we previously established. Three distinct DTriTE constructs were designed, varying the position of the CD3-binding fragment: centrally located in DT2035, at the N-terminus in DT1005, and at the C-terminus in DT3026 (figure 1A).

Figure 1

Design, in vitro expression, and on-cell binding of DTriTEs. (A) DTriTE configuration in pVAX1 vector: illustration showing DTriTE constructs within the pVAX1 vector and the three DTriTE variants with varied positioning of the CD3-binding scFv (central in DT2035, N-terminus in DT1005, C-terminus in DT3026) and the DBTE controls. (B) Western blot analysis: demonstrates the in vitro expression of DTriTEs and control constructs using transfected Expi293F cell supernatants. (C–D) Flow cytometry analysis of DTriTE binding: panels displaying the binding activities of DTriTE variants to (C) U87vIII, U251, and T cells, as well as to (D) DK, A172, U373, and LN18 cells. DBTE, DNA-encoded bispecific T-cell engager; EGFRvIII, epidermal growth factor receptor variant III; scFv, single-chain variable fragments.

In vitro expression was achieved by transfecting Expi293F cells with DTriTEs, followed by supernatant collection. As controls, we included our previously described DBTEs targeting EGFRvIII,10 IL-13Rα2,11 and FSHR,14 both as monospecific and irrelevant antibody controls, along with pVAX1 as a vehicle control. Western blot analysis of these supernatants revealed that DTriTEs are expressed with a higher molecular weight (~78 kDa) compared with the DBTEs (~52 kDa) (figure 1B).

Assessment of target tumor cell and DTriTE binding activities

To create a GBM model that encompasses the diverse expression of IL-13Rα2 and EGFRvIII, we employed the U251 cell line (IL-13Rα2+/EGFRvIII−) and U87vIII cell line (IL-13Rα2−/EGFRvIII+), which we developed by transducing U87 cells with EGFRvIII as previously described.10 Additional glioma cell lines were incorporated to represent varying expressions of EGFRvIII and IL-13Rα2, namely DK, A172, U373, and LN18. Flow analysis of these cell lines showed that U87vIII cells exhibited high EGFRvIII expression, U251 cells displayed high IL-13Rα2 expression, DK cells expressed both antigens, A172 cells also showed high IL-13Rα2 expression, while U373 and LN18 cells demonstrated low IL-13Rα2 expression (online supplemental figure 1).

To assess the binding activities of the DTriTE constructs, we added DTriTEs to the target cells and performed flow analysis. We observed that all variants of DTriTE showed robust binding activities to U87vIII, U251, and T cells (figure 1C). Notably, DT2035 exhibited a reduced binding activity to T cells compared with DT1005 and DT3026. The binding activities of all DTriTE variants were robust and comparable across DK, A172, U373, and LN18 cell lines, which exhibit varied expressions of EGFRvIII and IL-13Rα2 (figure 1D). The exception was observed with the A172 cell line, where DT1005 exhibited an increased binding than either DT2035 or DT3026 (figure 1D). Importantly, all DTriTE constructs showed binding to LN18 cells, characterized by low IL-13Rα2 expression levels and an absence of EGFRvIII.

T cell-mediated cytotoxicity of DTriTEs in EGFRvIII+ and IL-13Rα2+ tumor cells

We evaluated the ability of DTriTEs to induce T cell-mediated cytotoxicity using tumor-killing assays we previously developed.10 The glioma cell lines were cultured for 20–24 hours, after which DTriTE-containing supernatants and primary human T cells were added. Cell viability was continuously monitored using the xCELLigence system. We observed that all three DTriTE variants effectively induced cytotoxicity against EGFRvIII-expressing U87 and IL-13Rα2-expressing U251 cells (figure 2A,B). Microscopy of these cells revealed cell clusters indicative of effective cell death post-incubation with DTriTEs. Furthermore, DTriTE constructs demonstrated potent T cell-mediated cytotoxic effects on DK cells expressing both EGFRvIII and IL-13Rα2 (figure 2C), A172 cells expressing a high level of IL-13Rα2 (figure 2D), and even against U373 and LN18 cells which have low IL-13Rα2 expression (figure 2E,F). In contrast, the FSHR-DBTE, an irrelevant antibody control, did not show tumor-killing activity against any of the glioma cell lines tested (figure 2A–F). Additionally, when the DTriTEs were tested against non-glioma cell lines like ovarian (OVCAR8), pancreatic (Panc1), liver (Hep3B, HepG2) cancers, and kidney cells (HEK293T), no T cell-mediated cytotoxicity was observed (online supplemental figure 2).

Figure 2

DTriTE-induced T cell-Mediated cytotoxicity against glioblastoma multiforme cells. Cell viability of (A) U87vIII, (B) U251, (C) DK, (D) A172, (E) U373, and (F) LN18 were monitored using xCELLigence real-time cell analysis, followed by addition of DTriTE-containing supernatants (10 ng/mL) and primary human T cells (effector to target cell ratio of 10:1). Viability was normalized at the time of antibody addition to assess relative viability. The microscopic images of U87vII and U251 cells show the morphological changes in target cells at 0 and 24 hours post T-cell addition, with cell clusters indicating areas of T cell-induced cytotoxicity. DBTE, DNA-encoded bispecific T-cell engagers; DTriTE, DNA-encoded tri-specific T-cell engagers.

Additionally, to examine the potency of DTriTEs, we performed the tumor-killing assays against U87vIII and U251 cells at varying concentrations of DTriTEs and with different numbers of T cells. The EC50 values were determined as 2.35 pM for DT2035, 3.38 pM for DT1005, and 14.2 pM for DT3026 against U87vIII cells, while against U251 cells, the values were 32.0 pM, 8.18 pM, and 50.0 pM, respectively (online supplemental figure 3A,B). The cytotoxicity exhibited by all DTriTE variants ranged from 88% to 97% at effector-to-target ratios of 10:1 and 3:1, decreased to between 56% and 74% at a 1:1 ratio, and further reduced to between 10% and 25% at a 1:3 ratio, with no significant differences observed between the variants (online supplemental figure 3C). These findings demonstrate the potency of DTriTEs in promoting tumor cell death even at low concentration and at low effector-to-target cell ratios, highlighting their potential as an immunotherapy for GBM treatment.

T-cell activation by DTriTEs in antitumor responses

To explore the immunological effects of DTriTEs, we analyzed T-cell responses to DTriTEs. Following a 24-hour tumor-killing assay with DTriTEs, we isolated the T cells and performed flow analysis of the expressions of IFN-γ, TNF-α, IL-2 (antitumor cytokines), and CD107a (a marker for degranulation). We observed a significant upregulation of IFN-γ, TNF-α, and IL-2 in T cells treated with all DTriTE variants compared with vehicle control (figure 3A). In the killing assay against U87vIII cells, DT2035 elicited significantly higher IFN-γ, TNF-α, and IL-2 responses in T cells than EGFRvIII-DBTE, while DT1005 and DT3026 showed no difference. In the killing assay against U251 cells, DT3026 showed reduced T-cell responses in IFN-γ, TNF-α, and IL-2 than IL-13Rα2-DBTE, while DT2035 and DT1005 showed no difference. In comparison between DTriTE variants, we observed that DT2035 showed increased T-cell responses in IFN-γ and TNF-α than DT1005 and DT3026. DT2035 also showed better IL-2 response than DT3026. Importantly, we observed that all DTriTE variants did not show T-cell activation in the presence of OVCAR3 cells, an irrelevant target cell line, demonstrating the specificity of DTriTEs and the absence of off-target effects.

Figure 3

T-cell and NKT cell activation and cytokine release induced by DTriTEs. Flow cytometry analysis evaluating the activation of CD3+T cells following stimulation by DTriTEs, monovalent DBTEs, or pVAX1 in the presence of U87vIII, U251, or OVCAR3 cells. (A) Percentage of T cells showing increased expression of IFN-γ, TNF-α, and IL-2. (B) Percentage of T cells that are double-positive of IFN-γ+/TNF-α+ and triple-positive of IFN-γ+/TNF-α+/IL-2+. (C–D) IFN-γ, TNF-α, and IL-2 responses in (C) CD8+T cells and (D) CD4+T cells populations. (E) Percentage of CD107a-expressing CD8+and CD4+ T cells. (F) The proportion of CD3+CD56+ NKT cells exhibiting elevated levels of IFN-γ, TNF-α, IL-2, and CD107a. DBTE, DNA-encoded bispecific T-cell engagers; DTriTE, DNA-encoded tri-specific T-cell engagers; EGFRvIII, epidermal growth factor receptor variant III; IFN, interferon; IL, interleukin; NKT, natural killer T cell.

To determine the capability of DTriTEs to induce polyfunctional T-cell responses, we analyzed T cells secreting multiple cytokines. We observed that all DTriTE variants effectively generated IFN-γ/TNF-α double-positive and IFN-γ/TNF-α/IL-2 triple-positive T-cell populations compared with the vehicle control (figure 3B), indicating multifaceted immune responses by DTriTEs. Notably, DT2035 elicited higher levels of double-positive/triple-positive T cells than DT1005 and DT3026 in the presence of U87vIII cells.

To characterize these T-cell responses further, we analyzed CD8+ and CD4+ T-cell subpopulations in their responses to DTriTEs. We observed that DTriTEs induced significant activation of IFN-γ and TNF-α in CD8+T cells, but not of IL-2 (figure 3C). On the other hand, DTriTEs elicited significant activation of IL-2 in CD4+T cells (figure 3D). Notably, CD4+T cells also showed robust responses in the expression of IFN-γ and TNF-α, indicating that DTriTEs are capable of inducing cytotoxic activity in CD4+T cells in addition to CD8+T cells. These findings collectively suggest that DTriTEs can modulate the polyfunctional profile of T cells, potentially enhancing the efficacy of antitumor immune responses.

Additionally, we analyzed the activation of CD107a, a marker for T-cell degranulation, in T cells. We observed that all DTriTE variants promoted robust CD107a expression in CD8+T cells, but not in CD4+T cells (figure 3E). Notably, DT2035 induced significant activation of CD107a in CD4+T cells in the presence of U87vIII cells. To assess the secretion of perforin, granzyme B, and sFasL by T cells, we collected supernatant of the tumor-killing assay and analyzed for concentration of the granules. We observed that all DTriTE variants robustly promoted the secretion of perforin, granzyme B, and sFasL (online supplemental figure 4).

DTriTEs activate NKT cells

We sought to examine if DTriTEs can engage NKT cells which are a valuable target for immunotherapy due to their potent cytolytic activities and the capability to infiltrate tumors.15 Given their CD3 expression, NKT cells are particularly amenable to T cell-engaging immunotherapies. We conducted a tumor-killing assay using primary human T cells, selectively focusing on the CD3+CD56+ subset. Using the fluorescence-minus-one (FMO) method, we isolated a small percentage of the CD56+NKT cell subset to evaluate their activation which is on par with expected NKT levels12 (online supplemental figure 5). We observed that all DTriTE variants significantly enhanced the secretion of IFN-γ, TNF-α, and IL-2 in NKT cells when exposed to U87vIII and U251 cells (figure 3F). Additionally, all DTriTEs markedly increased CD107a expression in NKT cells (figure 3F). These findings indicate that DTriTEs effectively activate NKT cells, suggesting their potential role in enhancing antitumor immune responses within the tumor microenvironment through multiple effector populations.

In vivo expression of DTriTEs

To investigate the in vivo expression of DTriTEs, we intramuscularly injected 200 µg of DTriTEs into Balb/c nude mice, followed by electroporation. For comparison, we included groups of mice that received an intraperitoneal (IP) injection of recombinant tri-specific T-cell engager. Serum samples were longitudinally collected and used in a tumor-killing assay against U87vIII/U251 cells. A single injection of DTriTEs exhibited sustained in vivo expression evidenced by robust T cell-mediated cytotoxicity in the serum samples, effective for up to 105 days. In contrast, recombinant protein-based antibodies (rT2035, rT1005, and rT3026) showed peak cytotoxicity at day 1, which was dissipated by day 21 (figure 4A). Notably, DT2035 showed a prolonged and increased antitumor response compared with DT1005 and DT3026 (figure 4A). These findings demonstrate that DTriTEs maintain robust and sustained antitumor activity in a systemic model, with DT2035 particularly exhibiting prolonged and effective cytotoxic responses, highlighting its potential for long-term therapeutic applications.

Figure 4

DTriTE in vivo expression and activity in heterogeneous glioblastoma multiforme. (A) Balb/c nude mice (N=5) were given IM injection of 100 µg of DTriTEs or pVAX1, followed by electroporation, or IP injection of 100 µg of recombinant protein-based trispecific T-cell engagers (rT2035, rT1005, or rT3026). Serum samples were longitudinally collected and analyzed in a tumor-killing assay for assessment of in vivo expression and T cell-mediated cytotoxicity. (B) Tumor-killing assay demonstrating the efficacy of DTriTEs versus single-antigen-targeting DBTE controls against EGFRvIII/IL-13Rα2 heterogeneous tumor model, in which U87vIII and U251 cells were co-cultured at 1:1 ratio. (C) Comparative cytolysis percentages from the 48-hour tumor-killing assay from (B). (D) Fluorescent microscopy images captured from (B) at 0 and 12 hours post addition of effector cells and antibodies, illustrating target cells. Cell-tracing dyes differentiate U251 cells (in magenta) U87vIII cells (in cyan). DBTE, DNA-encoded bispecific T-cell engagers; DTriTE, DNA-encoded tri-specific T-cell engagers; EGFRvIII, epidermal growth factor receptor variant III; IM, intramuscular; IP, intraperitoneal.

Development of a heterogeneous GBM model

To establish a GBM tumor model that heterogeneously expresses EGFRvIII and IL-13Rα2, we co-cultured U87vIII-luc cells (EGFRvIII+/IL-13Rα2−) with U251-luc cells (EGFRvIII−/IL-13Rα2+) and conducted a T cell-mediated tumor-killing assay with DT2035. DT2035 demonstrated robust cytotoxicity against the mixed GBM cell population, whereas EGFRvIII-DBTE or IL-13Rα2-DBTE resulted in partial cytotoxicity (figure 4B,C). In contrast, FSHR-DBTE, an irrelevant antibody control, exhibited no cytotoxic effect (figure 4B,C). For visualization, U251 and U87vIII cells were stained with different fluorescent dyes. Fluorescent microscopy revealed that after a 12-hour incubation, DT2035 showed cytotoxicity in both tumor populations, while EGFRvIII-DBTE primarily affected U87vIII cells and IL-13Rα2-DBTE mainly targeted U251 cells (figure 4D). These results validate our GBM model’s heterogeneity and the necessity of targeting both EGFRvIII and IL-13Rα2.

Intracranial challenge of heterogeneous GBM in NOD scid gamma mice using DT2035

We have developed a rigorous orthotopic, heterogeneous GBM challenge model, using intracranial injection of U251-luc/U87vIII-luc cells in NOD scid gamma (NSG) mice. On sixth day following tumor inoculation, the mice were given a single intramuscular injection of DT2035 followed by an IP injection of primary human T cells the following day. Tumor progression was monitored using the in vivo imaging system (IVIS) spectrum. All mice treated with DT2035 survived, whereas no pVAX1-treated mice, only 50% of EGFRvIII-DBTE-treated, and 33% of IL-13Rα2-DBTE-treated mice survived (figure 5A). The mice treated with pVAX1 control showed no tumor impact, resulting in all mice succumbing to the tumor. EGFRvIII-DBTE-treated mice experienced partial tumor control, yet 50% eventually succumbed. IL-13Rα2-DBTE-treated mice initially controlled tumor growth, but this effect was transient, leading to 67% mortality. Conversely, DT2035-treated mice demonstrated rapid and effective tumor control within 8 days of treatment, resulting in survival for all DT2035-treated mice over the 35-day observation period (figure 5B,C). However, post-35 days, all surviving mice began to develop graft-versus-host disease (GvHD), unrelated to tumor growth, necessitating the termination of the study. This challenging intracranial GBM model in NSG mice demonstrates the superior efficacy of DT2035 in controlling aggressive tumor growth, achieving a significantly higher survival rate compared with treatments targeting single antigens. These results underscore the potential of DT2035 as a robust therapeutic strategy for heterogeneous GBM tumors, despite the onset of GvHD which is an intrinsic issue to the model, limiting the time of observation.

Figure 5

Orthotopic GBM challenges with EGFRvIII+/IL-13Rα2+ heterogeneous tumors. (A–C) Following intracranial inoculation of U87vIII-luc (3.3×104) and U251-luc (6.6×104) cells, NSG mice were administered 200 µg of DNA treatments (intramuscular injection) at day 6 and primary human T cells (1×107, intraperitoneal injection) at day 7. DNA treatments included DT2035 (N=8), EGFRvIII-DBTE (N=10), IL-13Rα2-DBTE (N=8), and pVAX1 (N=10). (A) Survival outcomes of the GBM challenge. (B) Tumor burdens measured by IVIS imaging. (C) IVIS images showing tumor burdens over the course of the challenge, with red crosses marking mice deceased or euthanized due to tumor progression. (D) In an extended challenge study, NSG-K mice were subjected to the same tumor challenge and treatment schedule, with additional DNA treatments on days 28 and 49. DNA treatments included DT2035 (N=9), a combination of EGFRvIII-DBTE and IL-13Rα2-DBTE (N=10), FSHR-DBTE (N=9), and pVAX1 (N=10). The graph shows the survival data from the long-term GBM challenge. DBTE, DNA-encoded bispecific T-cell engagers; EGFRvIII, epidermal growth factor receptor variant III; GBM, glioblastoma multiforme; NSG, NOD scid gamma.

Long-term survival of NSG-K mice in intracranial GBM challenge with DT2035

To investigate DTriTE’s long-term efficacy, we next moved to a second model and used NSG-K (major histocompatibility complex (MHC) I/II double-knock-out) mice, which should have lower GvHD responses. Following intracranial injection with heterogeneous GBM cells (U87vIII/U251), the mice were treated with DT2035 and primary human T cells, following the same regimens as the previous challenge. The mice were given additional DNA treatments on days 28 and 49. We monitored the mice for movement impairment and weight loss for survival. The results indicated a stark contrast in survival outcomes: all pVAX1-treated mice succumbed by day 24 and all FSHR-DBTE-treated mice by day 31 (figure 5D). Mice receiving the combined DBTE treatment showed initial improved survival, with half surviving until day 49, but none past day 66. Remarkably, nine of nine DT2035-treated mice survived up to 84 days, and six of nine persisted through 122 days of the challenge, showcasing DT2035’s potential for long-term tumor suppression in this model. All surviving mice did not show weight loss or adverse events (online supplemental figure 6).

RNA sequencing analysis of T-cell activation by DT2035

Next, we sought to understand the activation profile of T cells stimulated by DT2035 at the RNA level. We conducted a 24-hour tumor-killing assay against mixed U87vIII and U251 cells. We isolated T cells and performed RNA sequencing (RNA-seq). We observed that DT2035 significantly upregulated genes linked to antitumor activity, including granzyme B, IFN-γ, TNF-α, granzyme H, and FasL, while downregulating granzyme A, granzyme K, and TRAIL (figure 6A). Notably, genes related to T-cell differentiation, proliferation (IL-2, G0S2, IL-2Ra, CDKs, cyclins), and co-stimulation (4-1BB, OX40, CD40L, CD27, CD28) were elevated. Intriguingly, both immunosuppressive (IL-10, program cell death protein (PD)-1, cytotoxic T-lymphocyte associated protein (CTLA)-4, T-cell immunoglobulin and mucin domain-containing protein (TIM)-3, lymphocyte activation gene (LAG)-3, inducible T-cell co-stimulator (ICOS)) and immune cell trafficking genes (CCL3, CCL4) showed significant upregulation, while T cell immunoreceptor with Ig and ITIM domains (TIGIT) was downregulated. Ingenuity pathway analysis (IPA) of the significant genes identified activated regulators associated with T-cell proliferation (MYC, IL-15, IL-3, IL-2, IL-7, CSF2, PI3K, DR3), the T cell receptor (TCR) complex (CD3, TCR), and co-stimulation (CD28, 4-1BB, CD38, CD40L, OX40, IL-18, CD2), whereas type I IFNs and the regulatory T-cell marker FOXP3 were notably downregulated (figure 6B). This comprehensive RNA-seq analysis distinctly demonstrates that DT2035 not only enhances T-cell effector functions but also modulates a broad spectrum of immunoregulatory pathways, highlighting its potential as a multifaceted immunotherapeutic agent in GBM treatment.

Figure 6

RNA sequencing analysis of T-cell activation by DT2035. (A) Heatmap displaying significant differential gene expression (p<0.05) in CD3+T cells isolated post tumor-killing assay (24 hours) with DT2035 against U87vIII/U251 heterogeneous tumors. (B) Ingenuity pathway analysis identifying activated (red bars) and inhibited (blue bars) regulatory pathways in the T cells from the assay in (A).

DT2035’s efficacy in patient with engaging GBM PBMCs in antitumor activity

Finally, we investigated DT2035’s capacity to stimulate antitumor cytotoxicity in patient PBMCs. We conducted tumor-killing assays against heterogeneous cell mixture of U87vIII and U251 cells, adding PBMCs from five patients with primary GBM and DT2035. These PBMCs, graciously provided by Z Binder and D O’Rourke (the Perelman School of Medicine at the University of Pennsylvania), were from adult patients at various treatment stages. Patients A, B, and C were pretreatment, while patients D and E had undergone a regimen of radiotherapy and TMZ, with patient E subjected to more extensive treatments (online supplemental figure 7).

Our observations revealed that DT2035 consistently induced cytotoxicity in the PBMCs of all patients (figure 7A). The cytotoxic response was particularly pronounced in patient A. Further analysis involved examining the supernatants from these assays for cytokine release. We detected significant levels of antitumor cytokines (IFN-γ, TNF-α, IL-2) and granule proteins (perforin, granzyme B, sFasL) in the PBMCs from these patients, aligning closely with responses seen in healthy donors, particularly in patient A (figure 7B, online supplemental figure 7B).

Figure 7

Efficacy of patient with DT2035 GBM PBMCs. (A) Tumor-killing assays using PBMCs from patients with GBM A–C against U87vIII/U251 heterogeneous tumor cells. (B) Cytokine release assay results, displaying the levels of IFN-γ, TNF-α, IL-2, perforin, granzyme B, and sFasL in supernatant from the tumor-killing assays using PBMCs from patients A–C. The asterisks shown above bars indicate statistical significance in comparison to pVAX1 control. (C) Tumor-killing assays using PBMCs from patient D and patient E who received multiple rounds of chemotherapy and radiotherapy. GBM, glioblastoma multiforme; IFN, interferon; IL, interleukin; PBMC, peripheral blood mononuclear cell.

Most importantly, even after undergoing radiotherapy and chemotherapy, patients D and E’s PBMCs demonstrated significant cytotoxicity under DT2035 treatment (figure 7C). A series of microscopic images of the assay wells revealed the dynamic action of DTriTE-induced cytotoxicity by PBMCs of patient D and patient E (online supplemental figures 8A,B).

Additionally, it is critical to note that DT2035 did not induce notable responses in PBMCs against irrelevant targets like OVCAR3, highlighting its specificity in engaging effector cells.

Discussion

GBM remains a formidable therapeutic challenge due to its pronounced antigenic heterogeneity, which limits the effectiveness of monovalent therapies. This heterogeneity facilitates tumor escape, leading to poor patient outcomes.3 5 Our study introduces DTriTEs, a novel class of tri-specific T-cell engagers designed to address this complexity by targeting multiple tumor antigens, specifically EGFRvIII and IL-13Rα2, which are co-expressed in 76.7% and 51.2% of GBM cases, respectively, or for overall 93% of GBM cases when considered together.7 By engaging these antigens simultaneously, DTriTEs have the potential to mitigate tumor escape mechanisms and enhance therapeutic efficacy.

Our findings in this manuscript demonstrate the ability of DTriTEs to induce potent T cell-mediated cytotoxicity across a diverse range of GBM cell lines, including those with varying levels of antigen expression (figure 3). This is particularly significant given the resistance often observed in MGMT-unmethylated GBM, where standard therapies like TMZ fail due to robust DNA repair mechanisms within the tumor cells.16 The ability of DTriTEs to effectively target these difficult-to-treat cell lines, including LN-18, an MGMT-unmethylated cell line,17 18 underscores its potential as a versatile therapeutic agent capable of adding a new tool for impacting GBM’s inherent heterogeneity.

Importantly, DTriTEs demonstrated robust activation of T cells derived from patients with GBM, including T cells from patients who had undergone multiple rounds of chemoradiotherapy, which is a concern for impacting immune function19 (figure 7). The ability of DT2035 to engage and activate patient immune cells, leading to significant tumor cytotoxicity, suggests a translational relevance. These promising results suggest that DTriTEs may potentially elicit functional antitumor responses in patients whose T-cell immunity is weakened by prior treatments. Further studies using more diverse patient samples are necessary to elucidate factors that may influence the antitumor responses of DTriTEs and extend these observations.

Mechanistically, our flow cytometric study of DTriTE-stimulated T cells showed that DTriTEs activated both CD8+ and CD4+ T cells, inducing notable expression of CD107a, a marker of cytotoxic granulation (figure 3). This observation suggests that CD4+T cells, traditionally regarded as helpers in antitumor immunity, may also directly contribute to the cytotoxic activity against tumor cells when engaged by DTriTEs.20–22 The ability of DTriTEs to promote the generation of polyfunctional T cells, capable of secreting IFN-γ, TNF-α, and IL-2, further enhances its potential to drive a robust and sustained antitumor immune response (figure 3B). Such polyfunctional T cells play a pivotal role in effective antitumor immunity, as evidenced by a report that T cells producing TNF-α and/or IL-2 in addition to IFN-γ are abundantly present in the tumor-infiltrating lymphocyte population.23 Also, the specificity of DTriTEs in activating T cells against GBM cells, while sparing non-relevant cell lines, is crucial for minimizing off-target effects and maximizing therapeutic safety (figure 3A).

We also extend the potential therapeutic immune populations, demonstrating that NKT cell are also activated. The activation of NKT cells by DTriTEs is an important finding, as NKT cells play a crucial role in bridging innate and adaptive immunity, rapidly producing key cytokines like IFN-γ and TNF-α and exerting direct cytotoxic effects.24–27 Our study showed that DTriTEs robustly activated NKT cells, enhancing their cytotoxic function and cytokine production, which could amplify the overall antitumor response (figure 3F). This activation is particularly relevant for GBM treatment, as a higher presence of tumor-infiltrating NKT cells is associated with better clinical outcomes.28–30 By engaging NKT cells, DTriTEs not only target tumor cells more effectively but also potentially enhance the recruitment and activation of other immune cells, offering a broader and more sustained antitumor response, especially in tumors resistant to conventional therapies.

The RNA-seq analysis of T cells activated by DT2035 revealed significant immune modulation, revealing its enhanced antitumor activity at the gene level (figure 6). DT2035 upregulated key cytotoxic and immunomodulatory genes, promoting a more potent T-cell response. Notably, the upregulation of granzyme B and downregulation of granzyme A suggest that DT2035’s mechanism of action involves potent caspase-dependent cytotoxic pathways.31 Additionally, DT2035 promoted T-cell activation, proliferation, and co-stimulatory signaling, supporting overall robust T-cell engagement. The induction of immunoregulatory markers suggests potential synergy with checkpoint inhibitors, while the suppression of regulatory T cells points to enhanced antitumor immunity in potentially suppressive TME.32 Additional study of this new approach for translational studies is supported by these data.

The sustained in vivo expression of DTriTEs, as evidenced by prolonged antitumor activity in mouse sera following a single injection, represents a new extension over protein-based antibody treatments that require frequent dosing33 34 (figure 4A). This extended therapeutic window may suggest an enhanced overall efficacy of the treatment by maintaining continuous pressure on the tumor.

Our study evaluated three DTriTE variants, each with a distinct configuration of antigen-binding fragments: DT2035 with a central CD3-binding scFv, DT1005 at the N-terminus, and DT3026 at the C-terminus (figure 1A). DT2035 consistently outperformed the other variants, eliciting more polyfunctional T cells producing IFN-γ, TNF-α, and IL-2, (figure 3B) and demonstrating prolonged antitumor activity in mice (figure 4A). The central placement of the CD3-binding domain in DT2035 likely enhances its effectiveness by facilitating a balanced engagement of IL-13Rα2 and EGFRvIII, leading to a more effective immunological synapse.35 36 Additionally, DT2035’s lower binding affinity to T cells may enhance its specificity and efficacy in targeting tumor cells, potentially reducing T-cell activation-related toxicity by directing more antibody distribution to tumors rather than T cell-rich tissues37 38 (figure 1C). These results led us to select DT2035 for further evaluation in GBM challenge studies.

When compared with monovalent therapies, DTriTEs, particularly DT2035, demonstrated superior outcomes in both in vitro and in vivo models, achieving long-term survival and more comprehensive control of heterogeneous tumors (figure 5). These findings support the premise that multivalent targeting is essential for effectively addressing the antigenic diversity of GBM. The surgical GBM challenge in our study may have disrupted the blood-brain barrier, similar to surgeries performed in clinical settings, allowing T cells and DT2035 to infiltrate tumor sites and mount an antitumor response. Further study is necessary to elucidate the biodistribution of DTriTEs and T-cell trafficking in these models.

While promising, this study has several limitations. The in vitro and in vivo models, though effective, do not fully capture the complexity of the human tumor microenvironment, especially the immunosuppressive factors in patients with GBM. The long-term effects of sustained DTriTE expression, including potential immune-related toxicity, need further investigation. Reliance on immunocompromised mouse models limits our ability to assess full antitumor immune responses of DTriTEs. Additionally, the small patient sample size reduces the generalizability of our findings. Future studies should include a more diverse patient population and advanced models that better reflect human immune responses to thoroughly evaluate the safety and efficacy of DTriTEs.

The ability of DT2035 to activate patient-derived immune cells, particularly in the context of prior treatment history, suggests DTriTEs as a potentially important adjunct to existing treatment regimens. Future studies should focus on further optimizing the configuration of DTriTEs, exploring their synergistic potential with checkpoint inhibitors, and advancing their development into clinical trials. The potency of DT2035 in preclinical models supports immunotherapies that are multivalent are important to improve the treatment landscape for GBM, and other cancers characterized by high antigenic heterogeneity.

MethodsCell line generation and cultivation

The EGFRvIII-expressing U87-MG (U87vIII) cell line was established by transducing U87-MG tumor cells (HTB-14, ATCC) with a firefly luciferase lentivirus (PLV-10003, Cellomics Technology), followed by human EGFRvIII and green fluorescent protein (GFP) reporter complementary DNA. GFP-expressing cells were sorted, and EGFRvIII expression was confirmed using an anti-human EGFRvIII flow antibody (NBP2-50599, Novus Biologicals). The U251-luc cell line was created by transducing U251-MG tumor cells (09063001, MilliporeSigma) with a firefly luciferase lentiviral vector (PLV-10003, Cellomics Technology). DK-MG cells were procured from Amsbio (CL 01008-CLTH). U373-MG cells were obtained from MilliporeSigma (08061901). Additional cell lines including A-172 (CRL-1620), LN-18 (CRL-2610), PANC-1 (CRL-1469), 293T (CRL-3216), Hep 3B (HB-8064), Hep G2 (HB-8065), and OVCAR8 cells (courtesy of R Zhang, The Wistar Institute) were acquired from ATCC. All tumor cells were maintained at low passage numbers (below 20) and cultured in Minimum Essential Medium (MEM), Dulbecco's Modified Eagle Medium (DMEM), or Roswell Park Memorial Institute Medium (RPMI) 1640 enriched with 10% heat-inactivated fetal bovine serum (FBS) and 100 U/mL penicillin/streptomycin, at 37°C in a 5% CO2 atmosphere.

Expi293F cells, supplied by Thermo Fisher Scientific (A14527), were grown in Expi293 expression medium (A1435101, Thermo Fisher Scientific) and maintained in suspension using an orbital shaker, at 37°C in an 8% CO2 environment.

Primary human T-cell preparation

Primary human T cells, sourced from healthy donors at the Human Immunology Core at the University of Pennsylvania, were cultured in RPMI 1640 supplemented with 10% heat-inactivated FBS and 100 U/mL penicillin/streptomycin, at 37°C in a 5% CO2 incubator. For GBM challenge studies, T cells were activated and expanded using the T Cell Activation/Expansion Kit (130-09