Chronic antigen stimulation in melanoma induces T cell exhaustion and limits efficacy of T cell bispecific therapies

ABSTRACT

T cell bispecific antibodies (TCBs) have demonstrated promising results in patients with solid tumors, yet the immunological mechanisms influencing their efficacy require further investigation. T cell exhaustion, induced by prolonged antigen exposure, is known to compromise T cell-based immunotherapies, but its effect on TCB efficacy remains unclear. Herein, we assessed the TCB efficacy on tumor-specific T cells, emphasizing their functional status. Utilizing an immunocompetent mouse model with melanoma expressing an immunogenic antigen, we showed that tumor-specific T cells acquire an exhausted phenotype and fail to expand under TCB treatment. Both mouse and human tumor-specific T cells in vitro demonstrated that chronically stimulated T cells exhibit a reduced response to TCBs. The comparison of TCB efficacy in T cell-inflamed versus non-inflamed tumors in mice revealed TCB success depends more on T cell functional fitness than their initial abundance. These data underscore the importance of T cell exhaustion, suggesting that exhausted tumor-specific T cells are unlikely to be the primary effectors redirected by TCBs for tumor eradication. Our study highlights the need to maintain T cell fitness and prevent exhaustion to enhance TCB therapy outcomes, which may help identify patients who could benefit most from TCB treatments in clinics.

GRAPHICAL ABSTRACT

KEY POINTS

• The presence of an immunogenic tumor antigen results in increased infiltration of Regulatory T cells and CD8⁺ T cells expressing co-inhibitory receptors within the melanoma tumor.

• Chronic stimulation by tumor antigens induces exhaustion in tumor-specific T cells.

• Exhausted tumor-specific T cells exhibit impaired anti-tumor immunity when subjected to T cell bispecific treatment in melanoma.

• The efficacy of T cell bispecifics is likely determined by the functional phenotype of pre-existing T cells rather than the levels of T cell infiltration.

Introduction

Cancer immunotherapies have transformed the landscape of oncology by harnessing the cytotoxic potential of T cells to specifically target and eliminate cancer cells, delivering remarkable clinical outcomes.^1,2 The prevailing view in the field highlights the presence of tumor-specific cytotoxic T cells as a critical factor for achieving the therapeutic efficacy of immunotherapies.^3,4 However, a significant challenge remains: tumor-specific cytotoxic T cells frequently enter a state of exhaustion due to chronic antigen stimulation within the tumor microenvironment (TME), diminishing their anti-tumor functionality.⁵ This challenge underscores the need to investigate how T cells’ functional profile impacts the efficacy of newer therapeutic classes in cancer immunotherapy.

Over the past two decades numerous T cell-based immunotherapy strategies have been developed, broadly categorized into two subgroups: (i) those relying on cancer elimination via tumor-antigen recognition, such as checkpoint inhibitors (CPIs) which reinvigorate repressed T cells, and (ii) those whose specificity is controlled synthetically such as T cell-engaging bispecific antibodies (TCBs) that crosslink cancer cells with the patients’ T cells to facilitate killing, independently of their antigen specificity.^6–8

The tumor infiltration of antigen-specific T cells is initiated by the antigen-presenting cells (APCs) sampling and presenting tumor antigens to naive T cells in secondary lymphoid tissues. This process primes and activates tumor-antigen-specific T cells, enabling them to infiltrate and eradicate the tumor.⁹ Clinical trial findings suggest that tumor T cell infiltration serves as a prognostic marker for cancer immunotherapies.^10–13 However, a vast majority of patients with tumor-specific T cell infiltration still fail to achieve long-term responses to immunotherapies.² This failure can be attributed to T cell exhaustion resulting from persistent T cell receptor (TCR) stimulation, due to chronic exposure to tumor antigens,¹⁴ characterized by impaired cytotoxic capacity, reduced cytokine production, and elevated expression of inhibitory receptors.^15–17

Considering that continuous TCR stimulation might lead to an exhausted phenotype of antigen-specific T cells, it is crucial to prioritize alternative strategies such as TCBs that can also engage tumor-infiltrating “bystander” cells for anti-tumor immunity.¹⁸ TCBs are bi-specific antibodies with multiple-binding moieties recognizing a surface marker on tumor cells, while another binding moiety engages CD3ε on T cells, leading to subsequent T cell activation and tumor cell killing.^19–23 By delivering activation signals through CD3ε, TCBs operate independently of TCR-pMHC recognition,²⁴ recruiting T cells from the periphery and thereby demonstrating potential for efficacy in non-inflamed solid tumors.²⁵

Even though TCBs have been extensively studied,²⁶ their efficacy on tumor-specific cytotoxic T cells in solid tumors is not fully investigated yet. To address this gap, we explored (i) the efficacy of TCBs on tumor-specific T cells, (ii) the influence of the functional profile of these T cells on TCB efficacy, and (iii) the role of T cell fitness in therapeutic outcomes from TCBs in melanoma. Our study demonstrated that tumor-specific T cells do not expand by the TCB treatment in vivo, revealing for the first time that chronically stimulated tumor-specific T cells exhibit an impaired response to TCBs. Previously, the Glofitamab phase I trial observed that a high PD-1 gene expression signature on T cells correlates with reduced clinical response.²⁷ Similarly, the MajesTEC-1 study demonstrated that the response to TCB in multiple myeloma is linked to baseline immune cell fitness, with reduced responses observed in patients who have higher levels of T cells expressing co-inhibitory receptors and Regulatory T cells (T_regs).^28,29 In line with these findings, herein, we show that Tyrosinase-related protein 1 (Tyrp1) ×CD3 TCB also promotes a more favorable cytotoxic T cell phenotype in a murine melanoma model, characterized by lower levels of T_regs and reduced expression of inhibitory receptors on T cells at baseline.

Materials and methods

Therapeutic antibodies

All the therapeutic antibodies used in this work were produced internally in Roche.

Cell lines

OVA⁺ B16F10 FAP (originates from Reaction Biology) or B16F10 FAP (originates from ATCC) cells were cultured in DMEM GlutaMAX (Gibco, 10566016) with 10% fetal bovine serum (FBS) (Gibco, 26140079). The media for OVA⁺ B16F10 FAP cells was further supplemented with 1.5 µg/mL Puromycin (InvivoGen, ant-pr-1) and 400 µg/mL Hygromycin (Roche, 10566016), whereas B16F10 FAP cells were cultured in the presence of 0.6 µg/mL Puromycin (InvivoGen, ant-pr-1).

CHO-K1 cells (ATCC) transfected with a plasmid-encoding human TYRP1 were maintained in DMEM GlutaMAX (Gibco, 10566016), 10% FBS (Gibco, 26140079), 6 µg/mL Puromycin (InvivoGen, ant-pr-1).

Lenti-X™ 293T cells (Takara) and A375 cells (ATCC) were maintained in DMEM GlutaMAX (Gibco,10566016) with 10% FBS (Gibco, 26,140,079).

All cell cultures were incubated at 37°C in a humidified atmosphere containing 5% CO₂.

Killing assay with ex vivo generated OT-I T_eff and T_ex cells

Generation of OT-I T_eff and T_ex cells

OT-I T_eff and T_ex cells were generated following the previously described protocol.³⁰

Killing assay set up

OT-I T_eff and T_ex cells were labeled with 200 nM CellTrace™ CFSE Cell Proliferation dye (Thermo Fisher, 34554) for 10 minutes at 37°C. The 100 000 CFSE-labeled OT-I T_eff or T_ex cells were seeded into a 96-well U-bottom cell culture plates (TPP, 92697), in mouse T cell media comprising RPMI GlutaMAX (Gibco, 61870036), 10% FBS (Gibco, 26140079), 1% Sodium Pyruvate (Gibco, 11360039), 1% non-essential amino acids (Gibco, 11140035), 100 U/mL Penicillin-Streptomycin (Gibco, 15070063), and 0.05 mM β-mercaptoethanol (Gibco, 31350100). OVA⁺ B16F10 FAP or B16F10 FAP target cells were labeled with PKH-26 Red fluorescent dye (Thermo Fisher, PKH26GL) according to the manufacturer’s instructions. The 40 000 target cells/well were seeded into each corresponding well. Pre-diluted murine Tyrp1-TCB or control-TCB were added to the respective wells. Plates were incubated at 37°C in a humidified atmosphere containing 5% CO₂.

After 24–48 h of incubation, assay supernatant was either stored at −80°C for a cytokine release assay or used immediately to measure tumor cell lysis through a luminescence-based cytotoxicity assay. Target cells and OT-I cells were used for flow cytometry analysis.

In vivo experiments

The experimental study protocols were reviewed and approved by the Veterinary Department of Canton Zurich under the Veterinary License ZH180/2020 in agreement with The Swiss Animal Welfare Act. Experiments were conducted in compliance with the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC) ethical guidelines and reported in adherence to ARRIVE Guidelines 2.0. The number of animals used was kept to the minimum necessary to achieve statistically meaningful results. Animal housing, care, and experimental procedures followed the guidelines set by GV-Solas, Felasa, and TierschG.

In brief, female C57BL/6J mice, 6–8 weeks old, were purchased from Charles River. Mice were inoculated subcutaneously with 2 × 10⁵ OVA⁺ B16F10 FAP or B16F10 FAP cells, mixed 1:1 in Growth Factor-Reduced Matrigel (Corning, 356231) to a volume of 100 µL. Mice were maintained under specific-pathogen-free conditions according to guidelines (temperature of 22°C, dark/light cycle of 12 h and humidity of 50%; GV-SOLAS, FELASA), and food and water were provided ad libitum. Daily health monitoring was carried out. Tumor volumes were measured using a caliper, 1–3 times a week. Once tumor volumes were reached to 150–200 mm³, animals were randomized into treatment groups based on their tumor volume using an in-house generated automated software. Murine Tyrp1-TCB or vehicle control comprising Protein Buffer (Bichsel, 1000366) was administered intraperitoneal at a suboptimal dosage of 5 mg/kg weekly. Tumor growth inhibition was used as the readout. The animals were euthanized in case of tumor ulceration or once termination criterion was reached. In case of tumor ulceration, animals were excluded from the analysis.

Flow cytometry for in vivo samples

Single-cell suspension of tumors from in vivo experiments were stained with fluorophore-conjugated antibodies targeting desired surface and intracellular markers for flow cytometry analysis using the Symphony A3 (BD). Detailed protocol is provided in the supplementary methods.

Histology

Tumors collected from mice after sacrifice were fixed in 4% paraformaldehyde (Thermo Fischer, J60401-AK) overnight. The tumors were embedded in paraffin and were stained with either a standard H&E protocol using Mayer’s hematoxylin solution (Biosystems, 38702500) and eosin 2% (Biosystems, 84–0023-00), HLA-B (Abcam, ab240087) or with an anti-mouse CD3 antibody (Diagnostic Biosystems, RMAB005) following the manufacturer’s instructions. Images were captured with the slide scanner (Olympus, VS200).

Ex vivo killing assay

B16F10 FAP target cells were labeled with PKH-26 Red fluorescent dye (Thermo Fisher, PKH26GL) according to the manufacturer’s instructions. 5x10⁴ cells were seeded into a 96-well flat bottom plate (TPP, 92696) in previously described mouse T cell media.

CD8⁺ TILs were isolated from tumor single-cell suspensions using negative selection magnetic beads (Miltenyi, 130–116-478) and debris was removed with a debris removal solution (Miltenyi, 130–109-398). Isolated 1x10⁵ CD8⁺ TILs and pre-diluted TCBs were added into the corresponding wells. Plates were incubated at 37°C with 5% CO₂.

After 24–48 h of incubation, a part of the supernatant was either stored at −80°C for a cytokine release assay or used immediately to measure tumor cell lysis through a luminescence-based cytotoxicity assay. Target cells and TILs were stained for flow cytometry analysis.

Killing assay with MART-1 T_eff and T_ex cells

Generation and stimulation of MART-1 TCR T cells

MART-1 TCR T cells were generated by lentiviral transduction of CD8⁺ T cells isolated from buffy coats of healthy volunteers. The buffy coats were obtained from the Zürich Blood Donation Center, with informed written consent from donors and approval from the Cantonal Ethics Committee Zurich, in accordance with the Declaration of Helsinki. MART-1 TCR T cells were stimulated to generate MART-1 TCR T_eff and T_ex cells according to a previously published protocol, using acute or chronic tumor-antigen stimulation.³¹ Detailed protocol is provided in the supplementary methods.

Killing assay set up

CHOK1SV-TYRP-1 cells were harvested, washed, and seeded into 384-well V-bottom (Thermo Scientific, 4309) plates (40 000 cells/well). MART-1 TCR T cells, normalized to counts, were added to the CHO-K1 TYRP-1 cells at the same density. Pre-diluted human Tyrp1-TCB or control-TCB were added, reaching the desired concentration. The plates were incubated at 37°C in a humidified atmosphere with 5% CO₂. Part of the supernatant was either stored at −80°C for a cytokine release assay or used immediately to measure tumor cell lysis through a luminescence-based cytotoxicity assay. Subsequently, MART-1 TCR T cells were stained for flow cytometry analysis.

Cytokine measurement

Cytokines were analyzed from assay media using either BD™ Cytometric Bead Array (CBA) kits (BD), Bio-Plex Pro Mouse Cytokine 8-plex Assay (BIO-RAD, M60000007A) or LEGENDplex™ Human CD8/NK Panel (13-plex) w/VbP V02 kits according to manufacturer’s recommendation. Detailed protocol is provided in the supplementary methods.

Tumor cell killing – luminescence-based cytotoxicity assay

Promega Cytotox-Glo™ (Promega, G9291) was used to measure tumor cell lysis by detecting dead cell protein release. After 24–48 h of killing assay set up, the assay plates were centrifuged, and 20–50 µL of the assay supernatant was transferred to 384 well plates (Corning, 353988). Subsequently, Promega Cytotox-Glo™ reagent (Promega, G9291) was added to each well according to manufacturer’s recommendations. The plates were incubated (15 min, at RT, in dark) before the luminescence signal corresponding to the relative amount of lysed target cells, was measured using a Tecan SPARK 10 M (Tecan).

Flow cytometry for in vitro and ex vivo killing assays

Cells in the killing assay plates were washed twice with 70 µL/well PBS (Gibco, 20012027) before staining with 1:1000 diluted LIVE/DEAD™ Fixable Blue Dead Cell Stain (Invitrogen, L23105) at RT for 20 min. After washing the plate once more, cells were stained with a mix of antibodies, to surface markers (40 µL/well, 30 min, 4 C). Cells were washed with PBS (Gibco, 20012027) and fixed with Cytofix/Cytoperm™ Fixation/Permeabilization solution (BD, 554714) (40 µL/well, overnight, 4°C). After the fixation, cells were washed twice with Perm/Wash Buffer (BD, 554714) and stained with a mix of antibodies diluted in Perm/Wash Buffer (BD,554714) to intracellular markers Cells were washed once with Perm/Wash Buffer (BD, 554714) and diluted in 100 µL/well PBS (Gibco, 20012027) for acquisition using Symphony A3 flow cytometry (BD).

Receptor quantification

Receptors on MART-1 T cells were quantified at indicated time points using Quantum™ Simply Cellular® (Bangs Laboratories, 815B) kit according to manufacturer’s recommendation. Detailed protocol is provided in the supplementary methods.

Data analysis

Flow Cytometry data was analyzed using FlowJo V.10.10. GraphPad Prism V.9.5 was used to generate the graphs for statistical analysis. The statistical tests used are indicated in the figure legends.

Illustrations

Illustrations shown in this work were created using Biorender.com

Results

Immunogenic tumor-antigen induces a T cell-inflamed tumor microenvironment with tumor-specific T cells in vivo

To study TCB efficacy on tumor-specific T cells, we first confirmed the endogenous generation of tumor-specific CD8⁺ T cells and their activation status during tumor progression.

Immunocompetent mice were subcutaneously injected with either OVA⁺ B16 or B16 (OVA⁻ B16) melanoma cells. Tumor-infiltrating immune cells were analyzed at specific stages of tumor growth (Figure 1a). The presence of MHC-I-presented OVA immunogenic antigen with the capacity to induce endogenous immune response resulted in slower tumor progression (Figure 1b). It also led to the generation and infiltration of endogenously formed OVA-specific CD8⁺ T cells which were detected throughout different stages of tumor progression (Figure 1c).

Figure 1.

Presence of OVA antigen induces a T cell-inflamed tumor phenotype and generates activated antigen-specific T cells in vivo. (a) Schematic of the in vivo study design. (b) Tumor growth curves for subcutaneously injected B16 and OVA⁺ B16 tumors in syngeneic mice (n=16 per group; mean ± SEM). (c) Confirmation of endogenous OVA-specific CD8⁺ T cell infiltration of the tumor in the OVA⁺ B16 model on scout 1 (sc1), scout 2 (sc2) and termination (t) timepoints (mean ± SEM). (d) Expression levels of PD1, CD25 and Granzyme B in tumor-infiltrating OVA-specific versus non-specific CD8⁺ T cells in the OVA⁺ B16 model (medians indicated as horizontal lines). (e) Median abundance (%) of eight different immune cell populations (pie charts, gated on Live CD45⁺ cells) and quantification of tumor-infiltrating T cells (bar charts) in OVA⁺ B16 and B16 models at scout 1 (200 mm³), scout 2 (500 mm³), and termination (2000 mm³) time points. Each symbol represents a single animal, except in (b) where each dot represents n=8-16 animals. Vertical lines depict mean ± SEM. Statistical comparisons were performed using unpaired t-test with Welch’s correction, *p ≤ 0.05, **p ≤ 0.01.

Next, we assessed the activation and exhaustion status of OVA-specific CD8⁺ T cells. These cells displayed significantly higher levels of activation marker CD25 and inhibitory marker PD-1 compared to nonspecific CD8⁺ T cells, suggesting a highly stimulated phenotype (Figure 1d). Antigen-stimulated OVA-specific CD8⁺ T also exhibited higher baseline Granzyme B levels (Figure 1d). Furthermore, a decreasing trend in the Granzyme B effector marker was observed in both tumor antigen-specific and nonspecific CD8⁺ T cell compartments over the course of tumor progression (Figure 1d), potentially explaining tumor escape, despite the presence of OVA-specific CD8⁺ T infiltration.

As anticipated, presence of OVA antigen induced not only OVA-specific T cells infiltration but also a CD8⁺ and CD4⁺ T cell-inflamed TME (Figure 1e, Supplementary Figure S1a), along with increased infiltration of eight different detected intratumoral immune cell subsets (Supplementary Figure S1b). Furthermore, OVA⁺ B16 tumors had higher HLA-B expression levels confirmed by histology analysis (Supplementary Figure S2b).

TCB treatment does not provide significant tumor growth control or enhance the cytotoxic T cell phenotype in a T cell-inflamed melanoma model in vivo

Next, we aimed to evaluate whether tumors expressing immunogenic antigens, which are recognized by endogenous T cells, exhibit an improved therapeutic response to TCB treatment compared to tumors with little or no tumor antigen expression.

A murinized version of Tyrp1-TCB (muTyrp1-TCB) was selected as a model TCB, which targets Tyrp1 naturally expressed on the surface of B16 murine melanoma cells (Figure 2a).^32,33 To compare TCB efficacy in a T cell-inflamed tumor versus in a non-inflamed tumor model, C57BL/6 syngeneic mice were injected with either OVA⁺ B16 or B16 cells, respectively (Figure 2b).

Figure 2.

TCB treatment does not provide significant tumor growth control or enhance the cytotoxic T cell phenotype in OVA⁺ B16 model in vivo. (a) Graphical representation of muTyrp1-TCB design. (b) Schematic of the in vivo study design. (c) Tumor growth of subcutaneously injected OVA⁺ B16 or B16 cells treated with vehicle or muTyrp1-TCB shown till day 21, the last day before B16 vehicle group reached termination criteria. Treatment time points depicted with little arrows for each tumor model: red arrows for Vehicle/muTyrp1-TCB treatment for OVA⁺ B16 model, gray arrows for Vehicle/muTyrp1-TCB treatment for B16 model. (n=8-20 per group; mean ± SEM). (d) Ki67⁺ tumor-infiltrating CD8⁺ T cells upon two rounds of muTyrp1-TCB treatment in scout timepoint. (e) PD1⁺ TCF1^- effector CD8⁺ T cell levels (counts/mg and fold changes) in OVA⁺ B16 and B16 models treated with vehicle or muTyrp1-TCB (mean ± SEM). (f) Ratio of Tim3⁺ Granzyme B⁺ effector CD8⁺ T cells to Tim3⁺ Granzyme B^- exhausted CD8⁺ T cells (counts/mg and fold changes) in the tumor (mean ± SEM). (g) Regulatory T cell abundance (% with fold changes) in the TME (mean ± SEM). (h) Co-inhibitory receptor expression levels on tumor-infiltrating CD8⁺ T cells (% with fold changes) upon muTyrp1-TCB treatment over time (mean ± SEM). In (c), each dot represents at least 8 animals. Statistical comparisons were performed using non-parametric Mann-Whitney test, except for (c) where two-way ANOVA with Sidak test was applied, *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001.

In the OVA⁺ B16 model, despite the presence of high levels of tumor-infiltrating T cells at baseline (Figure 1e, Supplementary Figure S1a and Supplementary Figure S2a) and high Tyrp1 expression levels over the course of tumor progression (Supplementary Figure S2c), muTyrp1-TCB resulted in only moderate, non-significant tumor growth delay compared to the vehicle control group (Figure 2c and Supplementary Figure S3a). On the other hand, in mice injected with B16 tumors, despite low initial T cell infiltration levels (Figure 1e, Supplementary Figure S1a and Supplementary Figure 2a), muTyrp1-TCB treatment resulted in significant delay in tumor growth and improved survival (Figure 2c and Supplementary Figure S3a). While tumor growth control between both OVA⁺ B16 conditions and the Tyrp1 TCB treated B16 condition are similar, significantly higher tumor growth rates were observed for untreated B16 (Figure 2c). muTyrp1-TCB treatment of the B16 tumor model led to significant inhibition of tumor growth, whereas for the OVA⁺ B16 model, no such added tumor growth control beyond the baseline tumor antigen-specific response already present in the vehicle control group was observed (Figure 2c and Supplementary Figure S3a).

We also analyzed the phenotypic landscape of tumor-infiltrating CD8⁺ T cells in both tumor models. In the T cell-inflamed OVA⁺ B16 model treated with the vehicle control, a relatively high proliferative (Ki67⁺) and cytotoxic (Granzyme B⁺) CD8⁺ T cell infiltration was observed (Figure 2d and Supplementary Figure S3b). Consistent with the trend in tumor growth control, muTyrp1-TCB treatment did not result in a significant increase in these CD8⁺ T cell population in this model (Figure 2d and Supplementary Figure S3b). A similar pattern was also observed for effector (PD-1⁺ TCF-1⁻) CD8⁺ T cell infiltration (Figure 2e and Supplementary Figure 3c).

Conversely, in the non-inflamed B16 model, TCB treatment significantly enhanced the infiltration of proliferative (Ki67⁺), cytotoxic (Granzyme B⁺) and effector (PD-1⁺ TCF-1⁻) CD8⁺ T cells infiltration at the intermediate (scout) and termination timepoints (Figure 2d, Supplementary Figure S3b and Figure 2e). Additionally, muTyrp1-TCB treatment shifted the intratumoral better effector (Granzyme B⁺ Tim-3⁺)/exhausted (Granzyme B⁻ Tim-3⁺) CD8⁺ T cell ratio in favor of the highly cytotoxic better effector (Granzyme B⁺ Tim-3⁺) cells in the non-inflamed B16 tumors,³⁴ a shift not observed in OVA⁺ B16 tumors (Figure 2f and Supplementary Figure S3b). In line with previous reports,^35,36 muTyrp1-TCB treatment in both tumor models resulted in higher fractions of co-inhibitory receptors (PD-1 Tim-3 Lag-3) expressing CD8⁺ T cell and T_reg (FoxP3⁺ CD25⁺) (Figure 2g,h). Notably though, our data show that non-inflamed B16 tumors exhibited lower infiltration of these inhibitory T cells at baseline (Figure 2g,h). This observation aligns with previous reports on hematological tumors, indicating that patients with high T_reg infiltration and increased prevalence of exhausted T cell phenotypes at baseline exhibit a diminished response to TCB treatment.^28,29

Taken together, TCB treatment can effectively control tumor growth and promote a superior effector phenotype of tumor-infiltrating T cells in a non-inflamed melanoma model. We hypothesized that the TCB efficacy in the non-inflamed tumor settings was due to minimal tumor antigen presentation, which reduced the infiltration of inhibitory T cells, thereby providing TCB with a more responsive immune population to engage.

Tumor infiltrating CD8⁺ T cells from inflamed OVA⁺ B16 tumor fails to respond to TCB treatment ex vivo

To further assess the functional fitness of tumor-infiltrating CD8⁺ T cells (CD8⁺ TILs) ex vivo, T cells were isolated at the scout timepoint, after the second treatment round, from both tumor models and co-cultured with B16 cells (Figure 3a). Their ability to perform TCB-dependent target cell killing and secrete pro-inflammatory cytokines was evaluated (Figure 3b).

Figure 3.

Tumor infiltrating CD8⁺ T cells isolated from OVA⁺ B16 tumor fail to respond to TCB treatment ex vivo. (a) Graphical representation of ex vivo study protocol. (b) CD8⁺ TILs isolated from the scout timepoint (3 days after the 2^nd round of treatment) were re-stimulated either with muTyrp1-TCB or with muControl-TCB in the presence of B16 target cells for assessing their tumor cell lysis (24h) and cytokine production capacity (72h). (c) B16 cell lysis by CD8⁺ TILs upon muTyrp1-TCB engagement ex vivo, normalized to muControl-TCB (mean ± SEM). (d) IFN-ɣ and TNF--α production capacity of isolated CD8⁺ TILs upon muTyrp1-TCB engagement with B16 target cells (mean ± SEM). (e) Intracellular Granzyme B levels of isolated CD8⁺ TILs upon muTyrp1-TCB engagement with B16 target cells (mean ± SEM). Each dot represents the technical duplicates, which were generated as pooling samples of 4 experimental animals from the same tumor and treatment group. Vertical lines depict mean ± SEM.

Consistent with the in vivo findings, CD8⁺ TILs isolated from T cell-inflamed OVA⁺ B16 tumors, regardless of the prior in vivo treatments, failed to exhibit tumor cell killing ex vivo (Figure 3c). In contrast, CD8⁺ TILs isolated from TCB-treated B16 non-inflamed tumors demonstrated an enhanced tumor cell killing when engaged with muTyrp1-TCB ex vivo (Figure 3c).

Cytokine and Granyzme B production levels were also assessed. In alignment with the in vivo and ex vivo tumor cell lysis results, CD8⁺ TILs isolated from non-inflamed B16 tumors, irrespective of their in vivo treatment, produced IFN-γ and TNF-α upon muTyrp1-TCB restimulation (Figure 3d). In contrast, CD8⁺ TILs isolated from vehicle-treated T cell-inflamed OVA⁺ B16 tumors exhibited lower IFN-γ and TNF-α production, while those from TCB-treated OVA⁺ B16 tumors completely failed to produce cytokines upon restimulation (Figure 3d). CD8⁺ TILs isolated from both vehicle control and muTyrp1-TCB treated OVA⁺ B16 tumors showed elevated Granzyme B baseline levels in line with the in vivo observations (Supplementary Figure S3b), while ex vivo muTyrp1-TCB restimulation did not enhance these levels (Figure 3e), confirming ex vivo killing assay and in vivo findings (Supplementary Figure S3b,c). For the B16 model, in alignment with ex vivo killing assay (Figure 3c), CD8⁺ TILs treated with muTyrp1-TCB showed elevated baseline levels of Granzyme B, which showed a slight increase upon ex vivo muTyrp1-TCB restimulation (Figure 3e). CD8⁺ TILs from B16 exposed to vehicle control also show Granzyme B levels consistent with in vivo and ex vivo killing data (Supplementary Figure S3b,c), with low baseline values and no change upon ex vivo muTyrp1-TCB restimulation (Figure 3e).

These data confirm in vivo findings that CD8⁺ TILs from T cell-inflamed melanoma model with persistent antigen presence exhibit a functionally exhausted phenotype, characterized by impaired tumor cell killing and loss of polyfunctionality.

Antigen-specific CD8⁺ T cells do not expand upon TCB treatment in vivo

Next, we sought to evaluate the responsiveness of endogenous antigen-specific T cells to TCB treatment in vivo. To investigate this, syngeneic mice bearing OVA⁺ B16 tumor were treated weekly with muTyrp1-TCB or control vehicle. During tumor progression, tumors were analyzed for endogenous OVA-specific T cells (Figure 4a). Strikingly, muTyrp1-TCB did not increase intratumoral of OVA-specific CD8⁺ T cell levels (Figure 4b,c). Over the course of tumor progression, these OVA-specific CD8⁺ T cells acquired an effector (PD-1⁺ TCF-1⁻) phenotype (Figure 4d), followed by subsequent terminal exhaustion (Granzyme B⁻ Tim-3⁺), characterized by loss of Granzyme B expression (Figure 4e).³⁴ OVA-specific CD8⁺ T cells also expressed higher levels of co-inhibitory receptor (PD-1 Tim-3 Lag-3) levels associated with T cell exhaustion,^37–39 compared to nonspecific CD8⁺ T cells in the TME (Figure 4f).

Figure 4.

Antigen-specific CD8⁺ T cells do not expand upon Tyrp1 TCB treatment in vivo. (a) Schematic of the in vivo study design. Percentages (b) and amounts (c) of tumor-infiltrating OVA-specific CD8⁺ T cells at baseline, scout, and termination with vehicle or muTyrp1-TCB treatment (mean ± SEM). Infiltration of OVA-specific effector (d) and exhausted (e) CD8⁺ T cells in the tumor with vehicle or muTyrp1-TCB treatment over time. (f) Co-inhibitory receptor expression on OVA antigen-specific vs non-specific CD8⁺ T cells upon muTyrp1-TCB treatment over time. Each dot represents a single animal. Statistical comparisons were performed using non-parametric Mann-Whitney test, *p ≤ 0.05, **p ≤ 0.01.

Overall, this data suggests that chronically stimulated antigen-specific CD8⁺ T cells in the TME acquire exhausted phenotype and do not expand upon TCB treatment.

Exhausted mouse CD8⁺ T cells fail to provide anti-tumor immunity upon TCB engagement

Following the in vivo observation, we further assessed the impact of functional phenotype of antigen-specific T cells on TCB efficacy ex vivo.

For this, we repeatedly exposed OVA-specific murine OT-I CD8⁺ T cells to the OVA_(257–264) antigenic peptide ex vivo (Figure 5a). As previously reported,³⁰ this repetitive antigenic stimulation induces features of exhaustion in the OT-I CD8⁺ T cells (OT-I T_ex), thereby providing us a model for studying the efficacy of TCBs with exhausted OVA-specific T cells ex vivo. As a control, a separate group of OT-I CD8⁺ T cells was stimulated with the OVA _(257–264) peptide only once, to generate effector OVA-specific T cells (OT-I T_eff) (Figure 5a). In line with previous findings, OT-I T_ex cells showed significantly higher inhibitory receptor expressions (Supplementary Figure S4a-d). Upon re-encountering the antigen-presenting tumor cells (Supplementary Figure S4e), they showed reduced proliferation (Supplementary Figure S4f), production of pro-inflammatory cytokines (Supplementary Figure S4g) and tumor cell killing (Supplementary Figure S4h) compared to control OT-I T_eff cells.

Figure 5.

Exhausted OT-I CD8⁺ T cells exhibit impaired response to TCB treatment irrespective of antigen presence. (a) Graphical representation of the protocol used to generate activated OT-I T_eff and OT-I T_ex cells. (b) Mechanism of action of muTyrp1-TCB in the presence and absence of OVA antigen on the target B16 tumor cells. (c) Lysis of tumor cells by OT-I T_eff and T_ex cells following 24-hour TCB treatment. Background signals of tumor cells (dark grey), OT-I T_eff (blue) and OT-I T_ex (red) were shown as dotted lines. (d) Expression levels of CD25 in OT-I T_eff and T_ex cells after 72 hours of TCB engagement with tumor cells. (e) Expression levels of Granzyme B in OT-I T_eff and T_ex cells after 72 hours of TCB engagement with tumor cells. IFN-ɣ (f) and TNF-α (g) secretion levels in these T cells under the same conditions. Each symbol represents technical duplicates. Data are presented as mean ± SEM. Experiment was performed twice.

After concluding that generated OT-I T_ex cells resemble phenotypically exhausted tumor-specific T cells in vivo, OT-I T_eff and T_ex cells were co-cultured with Tyrp1-expressing B16 tumor cells in the presence of either muTyrp1- or a muControl-TCB, to evaluate these cells differential responses to TCB treatment ex vivo (Figure 5b). OT-I T_eff cells mediated strong dose-dependent tumor cell lysis upon muTyrp1-TCB engagement with less than 1 nM EC₅₀, whereas OT-I T_ex cells demonstrated reduced killing ability and required more than 7 fold higher TCB concentration (EC₅₀ = 7.6 nM), to mediate half maximal lysis (Figure 5c).

We further investigated the impact of chronic antigen exposure on TCB efficacy during simultaneous engagement with the tumor antigen. For this, OT-I T_eff and T_ex cells were co-cultured with OVA⁺ B16 target cells, expressing MHC-presented OVA antigen and cell surface receptor Tyrp1 as TCB target (Figure 5b). OT-I T_eff cells mediated strong tumor cell lysis upon OVA antigen recognition (Figure 5c), and the addition of Tyrp1-TCB could improve tumor cell killing only slightly (EC₅₀ = 0.15 nM) (Figure 5c). On the contrary, OT-I T_ex cells showed reduced tumor cell lysis at baseline; which can be attributed to the reduced TCR surface expression levels upon chronic antigen stimulation (Supplementary Figure S4i). They also failed to mediate TCB-dependent killing (EC₅₀ = 17.04 nM) (Figure 5c), suggesting that regardless of tumor-antigen presence, chronic antigen-stimulated T cells failed to perform TCB directed tumor cell killing.

Next, the expression of the late activation marker CD25 and the directed cytotoxicity marker Granzyme B on OT-I cells were evaluated. When co-incubated with B16 tumor cells OT-I T_eff cells, unlike OT-I T_ex cells, displayed a dose-dependent increase in CD25 and Granzyme B (Figure 5d,e). Notably, OT-I T_ex cells showed a higher baseline prevalence of Granzyme B (Figure 5e). On the contrary, when OT-I T_eff and T_ex cells were co-cultured with OVA⁺ B16 tumor cells, the functional capacity seemed to be already saturated by the recognition of the tumor antigen, as no significant differences were detected across different TCB concentrations (Figure 5d,e).

Cytokine secretion levels were also measured in the supernatants, revealing that OT-I T_eff cells produced higher levels of IFN-γ and TNF-α in dose-dependent manner, whereas OT-I T_ex cells failed to evoke a TCB-dependent cytokine response (Figure 5f,g). Notably, OT-I T_ex cells did exhibit some elevated IFN-γ levels upon antigen recognition on OVA⁺ B16 cells (Figure 5f,g), aligning with the intrinsic tumor cell killing observed when co-cultured with OVA⁺ B16 tumor cells (Figure 5c).

Overall, OT-I T_ex cells, due to persistent antigen stimulation, failed to mount an effective anti-tumor response upon TCB treatment. In contrast, OT-I T_eff cells, which were antigen-experienced but not chronically stimulated, demonstrated a strong TCB-mediated anti-tumor immune response. However, there was minimal additional TCB effect observed for tumor cells already presenting their specific antigen.

Chronic antigen-stimulated human MART-1 specific CD8⁺ T cells show impaired efficacy upon TCB treatment

After confirming that chronic antigen stimulation reduces responsiveness to TCB treatment both in vivo and ex vivo, we aimed to understand the impact of antigen-specific T cells’ functional phenotypes in a human context.

Given the limited availability of human tumor-specific T cells, human MART1-specific T cells were generated through transduction with a TCR sequence specific for the human melanoma antigen recognized by T cells 1 (MART-1). Following a previously published protocol,³¹ these engineered T cells were exposed to tumor cells presenting the MART-1 antigen under either acute or chronic conditions to derive effector (MART-1 TCR T_eff) or exhausted (MART-1 TCR T_ex) cells, respectively (Figure 6a). The functional status of these MART-1 TCR T_eff and T_ex cells was tested on MART-1 peptide presenting A375 tumor cells and confirmed by impaired tumor killing capacity of MART-1 TCR T_ex cells (Supplementary Figure S5a,b).

Figure 6.

Persistent antigen-stimulated human MART1-specific CD8⁺ T cells show impaired response to TCB treatment in vitro. (a) Graphical representation of the protocol used to generate MART-1 TCR CD8⁺ T_eff or T_ex cells³¹. (b) Cytolytic activity of generated MART-1 TCR T_eff and T_ex cells against CHOK1SV-TYRP1 tumor cells following 48 hours of huTyrp1-TCB treatment. (c) Granzyme B expression levels of MART-1 TCR T_eff or T_ex cells after 48 hours of huTyrp1-TCB engagement with tumor cells. Pro-inflammatory cytokines (d-e) and cytotoxic proteins (f-g) secretion levels of these T cells under the same conditions. Each symbol represents technical duplicates. Data are presented as mean ± SEM. Experiments were conducted using T cells isolated from two healthy donors.

To investigate the differential efficacy of TCBs on effector versus exhausted human antigen-specific T cells, MART-1 TCR T_eff and MART-1 TCR T_ex cells were co-cultured with CHOK1SV-TYRP1 cells expressing the human Tyrp1 receptor. As an additional control, mock TCR-transduced effector T cells (T_eff) from matching donors were also included. Then, cells were treated with varying concentrations of human Tyrp1 TCB (huTyrp1-TCB) or a human Control-TCB (huControl-TCB). In line with ex vivo mouse findings, MART-1 TCR T_eff cells mediated a strong dose-dependent tumor cell lysis upon huTyrp1-TCB engagement, similar to T_eff cells (Figure 6b). In contrast, MART-1 TCR T_ex cells exhibited a reduced tumor cell lysis capacity upon TCB treatment, failing to reach maximal lysis of T_eff and MART-1 TCR T_eff cells (Figure 6b).

The expression of the directed cytotoxicity marker Granzyme B was evaluated in these cells following huTyrp1-TCB treatment. Both donors’ T_eff cells and MART-1 TCR T_eff cells displayed a similar dose-dependent increase in Granzyme B expression levels (Figure 6c). In contrast, MART-1 TCR T_ex cells demonstrated reduced induction of Granzyme B expression upon huTyrp1-TCB engagement (Figure 6c). The secretion levels of cytokines and cytotoxic proteins, including IFN-γ, TNF-α, Perforin, and Granulysin, were measured in the supernatants (Figure 6d–g). Although MART-1 TCR T_eff cells were able to kill tumor cells as effectively as non-antigen-experienced T_eff cells from the same donors – mock TCR transduced but exposed to CD3/CD28 bead stimulation previously –⁴⁰ they exhibited slightly lower but TCB inducible levels of cytokine and cytotoxic protein secretion (Figure 6d–g). On the contrary, MART-1 TCR T_ex cells showed even lower levels of IFN-γ, TNF-α, Perforin, and Granulysin secretion levels upon huTyrp1-TCB treatment.

We also explored additional mechanisms triggered by chronic antigen stimulation which could contribute to the reduced efficacy of TCBs. In the ex vivo mouse setting, decreased surface TCR expression levels were observed on OT-I T_ex cells (Supplementary Figure S4i). This observation led us to hypothesize that lower TCR complex levels on chronically stimulated T cells might also result in decreased CD3 surface expression, potentially reducing CD3 availability for TCB engagement. To investigate this, TCR and CD3 levels on human MART-1 TCR T cells were quantified during the acute or chronic antigen stimulation, as well as during a subsequent resting period (Supplementary Figure S5a). Indeed, although both MART-1 TCR T_eff and T_ex cells initially exhibited high TCR/CD3 surface expression levels at baseline, they displayed reduced levels of TCR/CD3 following acute or chronic antigen exposure (Supplementary Figure S5c,d). Notably, acutely stimulated MART-1 TCR T_eff cells partially recovered their TCR/CD3 surface expression after the resting period, whereas chronically stimulated MART-1 TCR T_ex cells maintained lower TCR/CD3 expression levels (Supplementary Figure S5c,d).

This novel data set confirms that chronic antigen stimulation impairs TCB efficacy in human melanoma-specific T cells, highlighting the importance of T cells functional fitness for effective TCB treatments. It also suggests that reduced TCR/CD3 surface expression levels may explain the diminished efficacy of TCBs on chronically stimulated tumor-specific T cells.

Discussion

TCBs represent a promising class of cancer immunotherapy for solid tumors, offering significant therapeutic benefits to some patients, as demonstrated by the clinical use of the DLL3xCD3 TCB, Tarlatamab.^41,42 Recently, Friedrich et al.⁴³ showed that the activation of naive tumor-specific T cells is enhanced when these cells receive TCR-pMHC signaling simultaneously with TCB treatment. However, the field still lacks a comprehensive understanding of the potential influence of T cells’ phenotypic profiles on the therapeutic efficacy of TCBs in solid tumors.

Herein, we explored the efficacy of TCBs on antigen-specific T cells in melanoma, revealing that the functional phenotype of these T cells is a key determinant of their response to TCB treatment. Additionally, we showed that antigen-specific T cells were not expanded by TCB treatment in the used in vivo model as they acquired an exhausted phenotype in the TME. By investigating the TCB efficacy in the presence of tumor-associated antigen, we further showed that antigen presence can result in endogenous infiltration of the tumor with inhibitory T_reg and exhausted T cells, leading to only a transient response upon TCB treatment. Notably, we also highlight the TCBs potential to provide a significant anti-tumor response in non-inflamed tumors, enhancing cytotoxic effector T cells amount in the TME.

Clinical results from early cancer immunotherapy trials, particularly with CPIs, have established a consensus in the immuno-oncology field that tumors with immunogenic antigen presence, as evidenced by T cell infiltration, are more likely to benefit from these therapies.^4,44 Given that T cell-inflamed solid tumors represent only a small fraction of cancer cases,⁴⁵ we investigated how the efficacy of TCBs differ between T cell-inflamed and non-inflamed tumors, and whether pretreatment T cell infiltration is essential for TCB efficacy. Using OVA⁺ B16/B16 murine melanoma models treated with muTyrp1-TCB allowed us to compare TCB efficacy in T cell-inflamed versus non-inflamed TMEs side-by-side. Despite high pre-treatment T-cell infiltration, muTyrp1-TCB did not improve the tumor growth control in the T cell-inflamed OVA⁺ B16 model. This may be attributed to the elevated levels of inhibitory T_regs and exhausted T cells prior to TCB treatment. The non-inflamed model, lacking this baseline inhibitory T cell presence, exhibited improved tumor growth control following muTyrp1-TCB treatment. Our findings, consistent with previous clinical reports on hematological tumors, indicate that the functional fitness of preexisting T cells, rather than their infiltration levels prior to TCB treatment, is likely the primary determinant of potential TCB efficacy in melanoma.^28,29,43,46

In addition we also show how TCBs can be less efficacious in T cell-inflamed, immunogenic tumors, as the high level of immunogenicity can induce tumor-infiltrating T cells to adopt an exhausted phenotype. Taken together, these findings suggest potential benefits of using TCB treatments in non-inflamed solid tumors, where, despite the initially low numbers, T cells exhibiting a functional phenotype could be expanded within the tumor. However, it is also worth mentioning that mouse CD8⁺ T cells may respond differently to the TCB therapy compared to human CD8⁺ T cells, especially considering, the pathogen-free housing conditions of experimental animals that restrict the diversity and functional phenotype of tumor-infiltrating “bystander” T cells.^45,47

In solid tumors, tumor antigen-specific T cells are phenotypically and transcriptionally heterogeneous.^45,48 Therefore, assessing the impact of TCBs on these functionally diverse T cells in vivo is of significant importance. To address the question of how TCB treatment impacts antigen-specific T cells in the tumor, we focused on the OVA⁺ B16 tumor model, providing a traceable system to evaluate the efficacy of TCBs on endogenously generated, phenotypically heterogeneous OVA-specific CD8⁺ T cells. Aligning with clinical reporting, in T cell-inflamed OVA⁺ B16 model approximately 10 % of the total tumor-infiltrating CD8⁺ T cells were tumor antigen-specific,⁴⁸ and these cells exhibited a more exhausted phenotype compared to nonspecific bystander T cells. Supporting this finding, we observed that OVA-specific T cells within the tumors did not increase in numbers following muTyrp1-TCB treatment, suggesting they are not the primary drivers of the anti-tumor response induced by TCB treatment. However, it should be also noted that these data sets were produced using a highly immunogenic OVA antigen potentially amplifying the degree of exhaustion status in the tumor-infiltrating antigen-specific T cell populations, thereby enhancing their diminished response to the TCB treatment. Another question arising from this model is how nonspecific bystander CD8⁺ T cells respond to TCB treatment. Although these cells were not our primary focus, we hypothesize that the inhibitory TME, characterized by high levels of T_regs and exhausted T cells, contributed to functional impairment in the nonspecific bystander T cells, reducing their responsiveness to TCB treatment. Our hypothesis aligns with the observation’s of hematological tumors where TCB efficacy is lower in the prevalence of T_regs and exhausted T cells, despite the presence of bystander cells.^28,29,43,46 Future studies should focus on examining the response of both tumor-specific and nonspecific CD8 T⁺ cells to TCB treatment in other solid tumor models with physiological antigen presence.

Antigen-specific T cells represent only a subset of tumor-infiltrating cells, restricting the investigation of TCB efficacy on these cells. Through testing the muTyrp1-TCB on these ex vivo generated OT-I T cells, we showed that when exhausted, antigen-specific T cells do not respond to the TCB therapy. Our findings demonstrate that unlike effector phenotype OT-I cells, exhausted OT-I T cells can not contribute to anti-tumor immunity upon muTyrp1-TCB therapy. This ex vivo finding complements previous reports showing reduced TCB efficacy on exhausted T cells, suggesting that such diminished effectiveness can occur not only in hematological tumors but in solid tumors as well.^43,49

To address the limitations associated with using highly immunogenic tumor antigen and mouse models, we next employed human MART-1-specific CD8⁺ T cells, targeting a self melanoma antigen in human tumors.⁵⁰ Since antigen-specific T cells of patients constitute a fraction of tumor-infiltrating cells,⁴⁸ exhibit varying functional phenotypes,⁴⁵ and have different TCR affinities to the same antigen,⁵¹ we generated MART-1 TCR-transduced human CD8⁺ T cells. This approach enabled the generation of high numbers of MART-1-specific CD8⁺ T cells, all sharing the same TCR and baseline functional phenotype, with the potential to induce either an effector or exhausted phenotype.³¹ Consistent with in vivo observations, chronically antigen-stimulated human MART-1-specific cells exhibited an impaired response to huTyrp1-TCB treatment. As previous studies have shown that TCR-pMHC ligation leads to the ubiquitination and subsequent degradation of the TCR at the protein level,^52,53 and it is also known that, TCR:CD3 complex forms in the endoplasmic reticulum and gets transported to the cell surface as a unit;⁵⁴ in our study, we also assessed TCR:CD3 expression levels on MART-1 TCR T cells. Our preliminary data suggest that antigen encounter leads to a reduction in TCR:CD3 counts on both MART-1 TCR T_eff and T_ex cells. However, while acutely stimulated MART-1 TCR T_eff cells were able to restore their TCR:CD3 levels after resting in an antigen-free environment, chronically stimulated MART-1 TCR T_ex cells were unable to. Overall, our findings suggest that chronic antigen stimulation impairs antigen-specific CD8⁺ T cells in various mechanisms, leading to unresponsiveness to potent TCB treatment.

In this study, we revealed chronic antigen stimulation and subsequent T cell exhaustion as a potential mechanism contributing to resistance to TCB therapy in melanoma. Our data demonstrate that exhausted antigen-specific T cells, resulting from prolonged exposure to tumor antigens, likely fail to elicit an effective anti-tumor response or expand under TCB therapy. Our findings also indicate that the presence of immunogenic antigens can lead to a more exhausted TME, characterized by increased T_reg infiltration, accumulation of CD8⁺ T cells expressing co-inhibitory receptors, and reduced TCR:CD3 surface expression which, in turn, can reduce TCB efficacy in T cell-inflamed melanoma. Future plans include to investigate the combination of TCBs with co-stimulatory bispecific antibodies or targeted cytokines as a strategy to enhance TCB efficacy and overcome the exhaustion of tumor-specific T cells in solid tumors.

In light of our findings, we advocate for further investigation into the efficacy of different TCBs beyond melanoma, focusing on whether their effectiveness depends on pre-treatment T cell functional phenotype and fitness levels. Focusing on these aspects, rather than focusing solely on mutational burden or preexisting T cell levels, could enhance the therapeutic outcomes of TCBs, particularly in solid tumors which currently pose significant challenges in clinics.

Funding Statement

All funding for the studies were provided by Roche. The authors do not declare a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

Disclosure statement

All authors, except N. Joller and C. Münz are employees of Roche or were employed by Roche at the time of this study. All the authors, except I. HutterKarakoc, E.M. Varypataki, S. Lang and S. Simons, N. Joller, C. Münz declare ownership of Roche stock. All the authors, except N. Joller, C. Münz, have patents and royalties with Roche.

Data availability statement

The authors confirm that the data supporting the findings of this study are included within the article and the supplementary figures. Further information can be shared by the corresponding author, M. Amann, upon a reasonable request.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/2162402X.2025.2526444.

Abbreviations

APC

Antigen presenting cells

CPI

Checkpoint inhibitor

pMHC

Peptide- Major Histocompatibility Complex

TCB

T cell bispecific antibody

TCR

T cell receptor

TME

Tumor microenvironment

T_reg

Regulatory T cells

Tyrp1

Tyrosinase-related protein 1