The RCC cell lines 786-O (RRID: CVCL_1051) and ACHN (RRID: CVCL_1067), AML cell lines THP-1 (RRID: CVCL_0006) and MOLM-13 (RRID: CVCL_2119), and NSCLC cell line NCI-H1975 (RRID: CVCL_1511), were acquired from ATCC and cultured in RPMI medium (Corning, Manassas, VA) with 10% heat-inactivated fetal bovine serum (FBS; VWR, Radnor, PA). T-cell lymphoma cell lines, HuT 78 (RRID: CVCL_0337) and MJ (RRID: CVCL_1414), were acquired from ATCC and cultured in 80% IMDM (Corning) with 20% FBS. All cell lines were stably transduced with a lentivirus containing the luciferase gene and neomycin resistance and were selected with geneticin (Gibco, Grand Island, NY).

The C10 scFv clone was discovered by serial biopanning enrichment of a fully human naïve scFv phage display library (AbCheck, Czech Republic) against recombinant biotinylated human CD70 (Acro Bio, Japan), human IgG1 Fc-fused CD70 antigen (Acro Bio) and CHO cells expressing human CD70 (BPS Bioscience, San Diego, CA). Enriched clones were selected by flow cytometric staining of CD70-expressing CHO and/or human RERF-LC-KJ (RRID: CVCL_1654) cells and binding to recombinant His-tagged CD70 in an enzyme-linked immunosorbent assay (ELISA).

The 15F6 VHH clone was obtained by biopanning a phage display library generated from an alpaca immunized with a panel of human tumor cell lines, including KOPN-8 (RRID: CVCL_1866) and JVM-3 (RRID: CVCL_1320), that surface express human CD70. Anti-CD70 VHH clones were enriched during three rounds of immune phage library positive selection against recombinant human CD70 antigen, validated by fluorescence-activated cell sorting against human CD70 RCC cell lines and recombinant antigen via ELISA.

Unique enriched anti-CD70 clones were identified by HCDR3 clonotyping after Sanger sequencing and subcloned into a mammalian IgG1 Fc fusion expression plasmid for expression using an HEK system and purification using Protein A/G beads. Carterra LSA Surface Plasmon Resonance and OctetRed96 Biolayer Interferometry instruments were used to confirm CD70 binding competition against human CD27 and epitope bin grouping.

The CD70-targeted ϵ-TRuCs were generated by tethering the sequence of the CD70-specific scFv or VHH to the TCR CD3ϵ subunit via a flexible linker (GGGGSx3). ADP-520 TRuCs were generated by incorporating an mbIL-15 sequence into the anti-CD70 scFv (C10)-CD3ϵ construct following a 2A sequence. The mbIL-15 construct was generated by fusing human IL-15 to the full-length IL-15Rα sequence via a flexible linker. The constructs were cloned into lentiviral transfer plasmids for vector generation.

Lentiviruses were prepared by transient transfection of Expi293 suspension cells with packaging plasmids and the TRuC lentiviral transfer plasmids, as previously described (6). Cells were centrifuged 48 hours after transfection and supernatants were collected, filtered, and precipitated. After centrifugation, lentiviral pellets were resuspended in TexMACS medium (Miltenyi Biotec, Bergisch Gladbach, Germany) supplemented with 3% human AB serum (Gemini Bio, West Sacramento, CA) and aliquoted at -80°C until use.

Primary human T cells were isolated by magnetic bead separation using anti-CD4 and anti-CD8 microbeads (Miltenyi Biotec). On Day 0, T cells were activated using human T-Cell TransAct (40 μl per 1x10⁶ cells) and cultured in TexMACS medium with 3% human AB serum, 12.5 ng/mL human IL-7 (Miltenyi Biotec) and 12.5 ng/mL human IL-15 (Miltenyi Biotec). T cells were transduced with the respective lentiviral vectors on Day 1. Cells were expanded and supplemented with fresh media as appropriate and harvested on Day 10.

Stable luciferase-expressing tumor cells were plated in triplicates in a 96-well plate at 1x10⁴ cells per well in appropriate media for the tumor cell line used, and T cells were added at the indicated E:T ratios. After 24-hour culture, 50% of the culture supernatant was removed and preserved at -80°C for cytokine analysis. Tumor cell killing was determined using the Bright-Glo Luciferase Assay System (Promega, Madison, WI) according to the manufacturer’s protocol. Relative luminescence (RLU) was measured using the SpectraMax M5 plate reader (Molecular Devices, Sunnyvale, CA). The percentage of tumor lysis was calculated by the following formula:

Quantification of human cytokine levels was performed using a custom 4-plex Meso Scale Discovery (MSD) U-plex kit (Rockville, MD) according to the manufacturer’s instructions.

Donor-matched T cells were thawed and activated using TransAct microbeads (40 µL/1x10⁶ cells, Miltenyi Biotec) overnight. After 18–24 hours, activated T cells were washed to remove the microbeads and non-activated and activated T cells were labeled with 5 µM CTV (Invitrogen, Eugene, OR) in phosphate-buffered saline (PBS) for 10 minutes at 37˚C. The reaction was quenched by addition of AB serum (Gemini Bio). CTV-labeled resting or activated non-transduced (NT) T cells were resuspended in complete media (RPMI with 10% FBS) and co-cultured with anti-CD70 TRuC T cells (normalized to %TRuC⁺) or additional unlabeled resting NT T cells at indicated E:T ratios. The co-cultures were incubated at 37˚C, and after 48 hours, samples were evaluated by flow cytometry for non-TRuC T-cell loss. Killing of NT T cells was calculated using the following equation:

Human T cells were identified using the markers CD45 (HI30, BD Biosciences Cat# 563792, RRID: AB_2869519 or BioLegend Cat# 304014, RRID: AB_314402 or Cat# 304024, RRID: AB_493761), CD3 (OKT3, BioLegend Cat# 317322, RRID: AB_2561911; UCHT1, BioLegend Cat# 300460, RRID: AB_2564380), CD4 (RPA-T4, BioLegend Cat# 300526, RRID: AB_493743; SK3, BioLegend Cat# 344616, RRID: AB_2028483) and CD8 (SK1, BioLegend Cat# 344710, RRID: AB_2044010 or BioLegend Cat# 344714, RRID: AB_2044006 or BioLegend Cat# 344732, RRID: AB_2564624), and anti-CD70 TRuC⁺ cells were identified with anti-human IgG, F(ab’)₂ (Jackson ImmunoResearch Labs Cat# 109-545-006, RRID: AB_2337832). Additional markers were evaluated as indicated. Surface marker antibodies included: CD56 (Biolegend Cat# 318306, RRID: AB_604101), CD45RA (HI100, BioLegend Cat# 304140, RRID: AB_2563816), CCR7 (G043H7, BioLegend Cat# 353204, RRID: AB_10913813 or BioLegend Cat# 353208, RRID: AB_11203894), CD70 (113-16, BioLegend Cat# 355124, RRID: AB_2820006), CD69 (FN50, BioLegend Cat# 310912, RRID: AB_314847), PD-1 (EH12.2H7, BioLegend Cat# 329932, RRID: AB_2562256), LAG-3 (11C3C65, BioLegend Cat# 369314, RRID: AB_2629797), TIGIT (A15153G, BioLegend Cat# 372718, RRID: AB_2632933), CD27 (M-T271, BioLegend Cat# 356420, RRID: AB_2562603) and IL15Rα (JM7A4, BioLegend Cat# 330210, RRID: AB_2561440). Intracellular marker antibodies were evaluated as indicated including TCF1/7 (C63D9, Cell Signaling Technology Cat# 6709, RRID: AB_2797631 or Cell Signaling Technology Cat# 9066, RRID: AB_2797696) and Mcl-1 (LVUBKM, Thermo Fisher Scientific Cat# 51-9047-42, RRID: AB_2811847). Ex vivo evaluations were performed using anti-mouse CD45 (30-F11, BioLegend Cat# 103132, RRID: AB_893340; I3/2.3, BioLegend Cat# 147716, RRID: AB_2750449).

Flow cytometric staining was performed as follows: Cell samples were aliquoted into 96-well plates, washed with PBS and stained with LIVE/DEAD Fixable Blue viability dye (Invitrogen) diluted 1:1000 in PBS to distinguish live cells. Cells were washed with staining buffer (PBS with 2% FBS) followed by incubation for 20 minutes with 1:100 Anti-Human IgG, F(ab’)₂. Cells were washed again with staining buffer and resuspended in the remaining surface stain antibody cocktail. Flow cytometry was conducted on a BD LSRFortessa X-20 cytometer.

For the competition assay to detect CD70 on anti-CD70 TRuC T cells, C10 TRuC T cells were stained with increasing concentrations of phycoerythrin-labeled anti-15F6 hIgGFc and incubated at 4°C for 15 minutes. Cell samples were washed with staining buffer and fixed with 150 µL of FluoroFix (BioLegend) for 30 minutes. Fixative was washed out, and the samples were resuspended in staining buffer before performing flow cytometric analysis.

For intracellular staining, cells were fixed using the eBioscience Foxp3/Transcription Factor Staining Buffer Set (Invitrogen) according to the manufacturer’s instructions. Cells were then incubated with intracellular antibody cocktail for 30 minutes. Samples were washed and resuspended in staining buffer for flow cytometric analysis.

T cells were thawed and rested overnight at 37˚C in serum-free TexMACS media. After 18–24 hours, the cells were washed and plated at 1x10⁵ cells per well in a 96-well plate in either RPMI-1640 alone or supplemented with 10 ng/mL IL-15 or 100 nM ruxolitinib (Selleck Chemicals, Houston, TX). Cells were then directly fixed 1:1 with pre-warmed Cytofix Fixation Buffer (BD Biosciences) for 10 minutes at 37°C, washed with PBS with 2% FBS and permeabilized with pre-chilled Phosflow Buffer III (BD Biosciences) for 30 minutes at 4°C. pSTAT5 signaling was determined by staining with APC-labeled anti-human pSTAT5 (A17016B.Rec, BioLegend Cat# 936906, RRID: AB_2892500) for 30 minutes at room temperature.

T cells were plated at 1x10⁶ cells per well in a 48-well plate in 1 mL of complete media (RPMI-1640 with 10% FBS) alone or supplemented with either 100 IU/mL IL-2 or 10 ng/mL IL-15 and incubated at 37˚C. After 4 days, half of the cell suspension was removed for cell cytometric analysis to determine cell expansion and phenotype, and fresh media, either alone or supplemented with 100 IU/mL IL-2 or 10 ng/mL IL-15, was added to the wells and incubated at 37°C. This process was repeated every 3–4 days until Day 15. For evaluation of cell numbers via flow cytometry, ~5000 counting beads (123count eBeads, Invitrogen) were added to each sample prior to flow cytometric analysis.

T cells were co-cultured with target tumor cell lines at a 5:1 E:T ratio, with 1x10⁶ T cells and 0.2x10⁶ target tumor cells, in a 48-well plate in complete media alone or supplemented with 100 nM each of dasatinib, ruxolitinib, or tofacitinib (Selleck Chemicals) and incubated at 37°C. After 4 days, half of the cell suspension was removed for cell cytometric analysis to determine cell expansion and phenotype, and fresh media – supplemented with 100nM each of dasatinib, ruxolitinib or tofacitinib when indicated – was added back to the wells and incubated at 37°C. This process was repeated every 3–4 days until Day 15. For the evaluation of cell numbers via flow cytometry, ~5000 counting beads (123count eBeads) were added to each sample prior to flow cytometric analysis.

C10 TRuC T cells and NT cells were co-cultured with target tumor cells lines at 1:1 E:T ratio, with 3.5x10⁴ total or TRuC T cells and 3.5x10⁴ target tumor cells, in a 96-well plate in complete media and incubated at 37°C. The co-cultures were rechallenged every 4 days with 3.5x10⁴ fresh target tumor cells. Every 3–4 days until Day 15, the culture supernatants were harvested and stored at -80°C for later cytokine analysis by U-PLEX, and the co-cultures were harvested for flow cytometric analysis to determine T-cell expansion and phenotype. Separate conditions were set up for harvest at each collection timepoint. For the evaluation of cell numbers via flow cytometry, ~5000 counting beads (123count eBeads) were added to each sample prior to flow cytometric analysis.

Identical conditions were set-up in a 96-well white-walled plate to determine tumor cell lysis by assessing the luciferase activity from residual live tumor cells using the Bright-Glo Luciferase Assay System (Promega, Fitchburg, WI) according to the manufacturer’s instructions.

Bystander natural killer (NK) cell activation was measured in co-culture assays were performed using donor-matched primary human peripheral blood mononuclear cells (PBMCs). On Day 0, cryopreserved TRuC T cells (ADP-520, C10, or NT) and donor-matched PBMCs were thawed, washed, and counted. T cells were labeled with CTV and co-cultured with unlabeled PBMCs at a 1:1 ratio (1×10⁵ cells each) in complete media (RPMI-1640 with 10% of heat-inactivated FBS). Co-cultures were plated in 96-well round-bottom plates and incubated at 37°C, 5% CO₂ for 72 hours in the presence or absence of 786-O RCC target cells.

At endpoint, cells were stained for lineage (CD45, CD3, CD56) and activation (CD69) markers. Live/dead discrimination was performed using LIVE/DEAD Fixable Blue viability dye. NK cells were identified by flow cytometric analysis as CD3⁻CD56⁺CTV⁻, and CD69 expression was used as a surrogate marker for activation.

Recombinant human IL-15 (10 ng/mL; Miltenyi Biotec) was included as a positive control for IL-15–mediated NK cell activation. All conditions were run in technical triplicates, and ~5,000 123count eBeads were added to each well before acquisition for normalization of absolute cell counts to measure cell type fold-expansion by flow cytometric analysis.

Animals were handled under an Institutional Animal Care & Use Committee (IACUC) protocol managed by Charles River Accelerator and Development Lab and following the guidelines set forward by the Guide for the Care and Use of Laboratory Animals. Animal numbers per cohort were determined by power analysis, using standard deviations from kinetic models, an 80% power probability, and a 30% tumor volume reduction as the effect threshold (30). This analysis recommended four mice per group, but all studies included at least five mice per group.

6–8-week old female NSG (NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ; Strain# 005557, RRID: IMSR_JAX:005557) and NSG-MHC I/II DKO (NOD.Cg-Prkdc^scid H2-K1^tm1Bpe H2-Ab1^em1Mvw H2-D1^tm1Bpe Il2rg^tm1Wjl/SzJ; Strain# 025216, RRID: IMSR_JAX:025216) mice were acquired from Jackson Laboratory (Bar Harbor, ME), and all animal studies were approved by the Charles River Laboratories Animal Care and Use Committees. For the systemic MOLM-13 AML model, 5x10⁴ MOLM-13-Luc cells were infused intravenously into the tail vein of NSG mice. Four days post-tumor injection, mice were randomized into treatment groups and infused with 10⁷ C10 TRuC, 15F6 TRuC or NT T cells. For the 786-O model, 3x10⁶ 786-O-Luc cells were implanted subcutaneously into NSG or NSG-MHCI/II DKO mice. Eighteen days post-tumor implant, tumor sizes reached ~150–200mm³, and mice were infused intravenously with 0.5–3x10⁶ anti-CD70 TRuC or NT T cells. For the ACHN model, 2x10⁶ ACHN-Luc cells were implanted subcutaneously into NSG-MHCI/II DKO mice and allowed to engraft for ~12 days to reach a tumor size of 150–200mm³. Randomized mice were infused intravenously with 1x10⁶ TRuC T cells or NT T cells. Across treatment groups, TRuC T cells were normalized to %TRuC⁺ by the addition of NT T cells as necessary. Tumor burden was determined by caliper measurements (Tumor volume = (length × width²)/2) and/or IVIS bioluminescence imaging (PerkinElmer, Waltham, MA) twice per week. Euthanasia was performed by trained staff and accomplished using CO₂ asphyxiation followed by cervical dislocation, in accordance with our IACUC protocol.

For the ex vivo evaluation of TRuC T cells, submandibular blood was collected into ethylenediaminetetraacetic acid tubes and mice were euthanized for tumor excision at indicated timepoints. Blood samples were subjected to red blood cell lysis using ACK lysis buffer (Gibco) and washed with PBS to remove clotted lysed cells. Subcutaneous tumors were enzymatically digested using the Tumor Dissociation Kit, human (Miltenyi Biotec) and filtered to obtain a single cell suspension. Single cell blood and tumor cell suspensions were aliquoted and analyzed by flow cytometry to evaluate TRuC T-cell infiltration.

GraphPad Prism (RRID: SCR_002798) was used for statistical analyses. In vitro experiments were performed with three technical replicates. Treatment groups were compared via one-way or two-way analysis of variance (ANOVA), followed by Dunnett’s multiple comparison test, as appropriate. Ex vivo experiments were performed with three biological replicates, and treatment groups were compared using one-way ANOVA with Dunnett’s multiple comparison test. For in vivo studies, treatment groups were compared using a two-way repeated-measures ANOVA, followed by the Bonferroni post hoc test. Statistical significance was systematically defined as P<0.05.

We executed three strategies to generate high-affinity anti-CD70 antibody fragments for assimilation into the TRuC T-cell platform, including evaluation of variable heavy domain of heavy chain [VHH] (Track 1), fully human and naïve single chain variable fragment [scFv] (Track 2), and humanized hybridoma-derived scFv (Track 3) libraries ( Supplementary Figure 1A ). These campaigns generated an enriched CD70 antibody fragment library comprising 112 unique heavy-chain complementarity-determining region 3 (HCDR3) clonotypes that exhibited an affinity range of <1–56 nM (K_D) for CD70, competing with CD27 for CD70 binding. Of these, 25 antibody fragments were formatted into and screened in the TRuC T-cell platform via N-terminal fusion to full-length human CD3ε.

Healthy donor CD4⁺ and CD8⁺ T cells were transduced using lentiviral vectors that encoded the candidate anti-CD70 TRuCs. TRuC T cells generated by fusion with Track 1 or Track 3 anti-CD70 binders exhibited CD70-dependent fratricide, characterized by poor T cell expansion, high CD69 expression, and terminally differentiated memory phenotype, with Track 3 also showing poor in vitro effector functions (properties summarized in Supplementary Figure 1A ). TRuCs constituting anti-CD70 fragments derived from Track 2 and formatted as N-terminal fusion of variable heavy chain to variable light chain (vHvL), or vice versa (vLvH), yielded several candidates with >10-fold expansion ( Supplementary Figure 1B ) and/or TRuC⁺ surface staining above the >50% cutoff ( Supplementary Figure 1C ). Half of Track 2 TRuC T cells exhibited a CD70^low/CD69^hi expression pattern that portends fratricidal activity, while one TRuC T-cell product showed a CD70^hi/CD69^hi expression pattern suggestive of target-independent auto-activation ( Supplementary Figure 1D ). While several Track 2 TRuC T cells exhibited markers of fratricide and auto-activation and were eliminated from downstream development, a subset remained CD69^low and maintained a memory phenotype dominated by naïve and central memory populations ( Supplementary Figure 1E ), as ascertained through expression levels of CD45RA and chemokine receptor 7 (CCR7). Together, these data identified several Track 2-derived TRuC T-cell candidates with favorable fratricide-resistant phenotypes for further testing.

The in vitro potency of CD70 TRuC T cells was assessed by measuring their tumor lytic activity and inflammatory IL-2 and interferon-gamma (IFN-γ) cytokine release response to CD70⁺ tumors. TRuC T cells were titrated by E:T against luciferase-expressing tumor cell lines, THP1, ACHN and 786-O, exhibiting a range of CD70 expression levels ( Supplementary Figures 1F–H ). While five TRuC T-cell products efficiently killed all three tumor cell lines, the Track 2 C10 vLvH TRuC T cells secreted markedly higher levels of both IFN-γ and IL-2 in response to tumor culture compared with all other candidates. Overall, the C10 vLvH TRuC T-cell product demonstrated a fratricide-resistant phenotype (favorable expansion, high TRuC expression, favorable memory phenotype distribution, low tonic activation) and high on-target effector function ( Supplementary Figure 1I ). The superiority of the C10 vLvH TRuC T-cell product, hereafter referred to as C10 TRuC T cells, led to its selection as a lead candidate for further characterization and development.

We further investigated the ability of C10 TRuC T cells to avert bystander cytotoxicity by directly measuring their effector function response against CD70-expressing T cells compared with a fratricidal TRuC utilizing Track 1 anti-CD70 clone 15F6 (see experimental design, Figure 1A ). As expected, activation of T cells with anti-CD3/anti-CD28 beads resulted in upregulated expression of CD70 in both CD4⁺ and CD8⁺ T cells ( Figure 1B ). Resting or activated T cells were labeled with CellTrace Violet (CTV), co-cultured with CD70-targeted TRuC T cells for 48 hours and evaluated for killing by flow cytometry. C10 TRuC T cells did not show significant killing of either resting or activated CD70-expressing CTV⁺ NT T cells ( Figure 1C ; Supplementary Figure 2A ). In contrast, a TRuC product known to exhibit fratricide, clone 15F6, showed a low level of killing of resting T cells (~12%) and ~36% killing of activated T cells after 48 hours of co-culture ( Figure 1C ; Supplementary Figure 2A ), comparable to the frequency of CD70⁺ expression present in the respective T cell populations ( Figure 1B ).

Studies of CD70 CAR T cells have previously indicated that a CD70-CAR cis interaction may occur on the CAR-T cell surface, masking the detection of CD70 and protecting the T-cell therapy from fratricide (29). Similarly, CD70 is not detected on C10 TRuC T cells ( Supplementary Figure 1D ). To evaluate whether a similar mechanism was occurring with C10 TRuC T cells, we performed a competition assay by titrating the addition of a fluorochrome-conjugated 15F6 binder, which shares an epitope bin with the C10 TRuC and therefore competes for CD70 binding on activated C10 TRuC T cells. Increasing concentrations of 15F6 resulted in the detection of CD70 on C10 TRuC T cells ( Supplementary Figure 2B ), suggesting that reduced detectability of CD70 on C10 TRuCs may be due to C10 TRuC-mediated cis masking of surface-expressed CD70, rather than CD70 depletion due to fratricide. Overall, these data confirm that C10 is a potent antitumor CD70 TRuC T-cell product that does not show significant fratricide or killing of normal bystander CD70⁺ T cells, at least partly due to the masking of surface CD70 by the TRuC itself.

We next evaluated the in vivo antitumor efficacy of C10 compared with 15F6. In NOD SCID gamma (NSG) mice infused with the AML cell line MOLM-13-Luc, treatment with either C10 or 15F6 TRuC T cells effected a comparable and complete MOLM-13 tumor clearance that persisted to 50 days post-treatment ( Figure 1D ). Conversely, in a subcutaneous RCC (786-O) mouse model, C10 TRuC T cells induced potent and rapid tumor regression in all mice whereas 15F6 only delayed tumor growth and did not achieve full clearance in 4/5 mice ( Figure 1E ). The C10 TRuC T-cell-treated mice that showed complete RCC tumor regression were re-challenged with 786-O tumor cells on the opposite flank 43 days after treatment. Compared with naïve mice, which all exhibited 786-O tumor growth, only one of four re-challenged mice in the C10 TRuC T-cell-treated group showed tumor growth, with the prevailing 75% of other animals remaining tumor free ( Figure 1F ). This finding indicates that fratricide-resistant C10 TRuC T cells drive potent and persistent antitumor efficacy in vivo.

Although C10 TRuC T cells demonstrated favorable characteristics relative to other screened candidates ( Supplementary Figure 1 ) – including fratricide resistance and strong on-target effector function – several features were noted that might limit long-term persistence, including modestly elevated CD69 expression ( Supplementary Figure 1D ) and a trend towards increased effector memory cells re-expressing CD45RA (T_EMRA) and reduced naïve cells ( Supplementary Figure 1E ). Utilizing IL-15 signaling to improve TRuC T-cell persistence, we generated ADP-520 TRuC T cells by combing the fratricide-resistant C10 TRuC receptor with membrane-bound IL-15:IL-15Rα fusion protein (mbIL-15) ( Figure 2A ). ADP-520 showed similar expansion during the TRuC T-cell product generation process ( Figure 2B ) and high transduction efficiency (~60–70% TRuC⁺) comparable to C10 TRuC T cells (80–90% TRuC⁺) ( Figure 2C ). ADP-520 showed masked CD70 expression that was comparable to C10 TRuC T cells ( Figure 2D ), and the addition of mbIL-15 did not increase target-independent auto-activation, as baseline levels of CD69 expression remained low and comparable to C10 TRuC T cells ( Figure 2E ). C10 TRuC T cells themselves showed a modest increase in terminally differentiated effector (T_EMRA, CD45RA⁺CCR7^-) cells ( Figure 2F ) and CD69 expression under monoculture conditions ( Supplementary Figure 1D ) compared to NT cells, possibly suggesting a low-level tonic C10 TRuC activation. However, these findings are not indicative of overt antigen-independent tonic signaling and instead demonstrate the low potential of C10 TRuC to elicit ligand-independent autoactivation. Comparatively, ADP-520 showed reduced T_EMRA CD4⁺ and CD8⁺ populations and an increased proportion of CD8⁺ naïve/stem cell memory (T_SCM)-like (CD45RA⁺CCR7⁺) T-cell populations ( Figure 2F ). Since C10 and ADP-520 TRuCs were generated using identical media containing exogenous IL-15, these differences may under-estimate that impact of mbIL-15 co-expression on memory phenotype distribution and reflect the cis-signaling activity of mbIL-15, rather than differences in media formulation. ADP-520 also maintained the fratricide resistance of C10 TRuC T cells and exhibited little effect on the number of resting or activated CTV⁺ T cells ( Supplementary Figure 2C ).

Given that the mbIL-15 induces a reduced effector population and a higher proportion of naïve T cells within ADP-520 ( Figure 2F ), we next evaluated the potency of ADP-520 against CD70-expressing RCC cell lines, 786-O and ACHN. Compared with C10 TRuC T cells, ADP-520 displayed robust target-dependent cytokine production and antitumor activity against both the CD70^hi 786-O and CD70^mod ACHN cell lines, but not CD70^neg K562 ( Figures 3A–C ) or CD70-knockout (KO) 786-O cell lines ( Supplementary Figure 3A ). However, ADP-520 demonstrated higher IL-2 levels in response to CD70^hi 786-O ( Figure 3A ). Further, ADP-520 demonstrated robust cytotoxicity and cytokine production against additional CD70-expressing cell lines, including the MJ ( Supplementary Figure 3B ) and HuT78 ( Supplementary Figure 3C ) T-cell lymphoma cell lines, and the non-small cell lung cancer (NSCLC) line, NCI-H1975 ( Supplementary Figure 3D ). Collectively, these data show that ADP-520 TRuC T cells exhibit reduced signs of autoactivation and terminal differentiation while retaining strong efficacy against CD70-expressing target cells of multiple tumor indications.

IL-15 signals through the Janus kinase (JAK)1/3-signal transducer and activator of transcription (STAT) 5 pathway in T cells (17) and is known to promote the transcription of B-cell lymphoma 2 (BCL-2)-family genes, including anti-apoptotic induced myeloid leukemia cell differentiation protein 1 (MCL1) (31). To test whether the mbIL-15 enhancement engages IL-15 signaling pathways, we investigated the phosphorylation status of STAT5. TRuC T cells were rested overnight in serum and cytokine-free medium before harvesting for phospho-STAT5 (pSTAT5) detection. ADP-520 TRuC T cells showed increased resting pSTAT5 baseline signal compared with the C10 TRuC T cells and NT T cells ( Figure 4A ). In accordance, the ADP-520 TRuC T cells showed increased expression of MCL-1, particularly in CD8⁺ T cells ( Figure 4B ). Moreover, treatment of resting ADP-520 T cells with FDA-approved JAK1/2 inhibitor ruxolitinib reduced elevated STAT5 phosphorylation to levels comparable to untreated C10 TRuC T cells, suggesting that mbIL-15 confers higher basal IL-15 signaling to ADP-520 T cells ( Supplementary Figure 4A ). The reported ability of IL-15 to promote and maintain a T_SCM population capable of self-renewal (17) was supported by data that ADP-520 displays increased CD45RA⁺CCR7⁺ naïve T cells ( Figure 2F ). To further understand the impact of mbIL-15 on TRuC T-cell stemness, we evaluated the expression of the transcription factor, T cell factor 1 (TCF1, encoded by TCF7), an important regulator of T-cell development and differentiation that, in conjunction with CD27, marks quiescent T-cell populations that are capable of increased self-renewal (32, 33). ADP-520 TRuC T cells demonstrated an increased frequency of expression of both TCF1 and CD27 ( Figure 4C ) compared with C10 or NT T cells, indicating that stemness and self-renewal potential are increased by mbIL-15 in ADP-520 TRuC T cells.

To evaluate whether IL-15 expression in ADP-520 induces bystander NK cell expansion – a potential source of IL-15 associated toxicity – TRuC T cells were co-cultured with donor-matched PBMCs in the presence or absence of CD70^hi 786-O tumor cells ( Supplementary Figure 4B ). After 24 hours of co-culture, flow cytometry analysis revealed no increase in CD3^-CD56⁺ NK cell frequency and negligible NK cell activation, as indicated by CD69 induction, in co-cultures with either ADP-520 or C10 TRuC T cells ( Supplementary Figure 4C ), unlike the robust NK cell activation observed upon treatment with recombinant IL-15 (positive control). These results indicate that mbIL-15 ADP-520 does not drive potent NK cell transactivation or proliferation, supporting its activity as a cis activating enhancement in ADP-520 with small impact on bystander NK cells and low risk of NK cell-associated toxicity.

To further explore the impact of the mbIL-15 enhancement on TRuC T-cell persistence, we cultured TRuC T cells in the absence of antigen stimulation or exogenous cytokines. In contrast to C10 TRuC and NT T cells, which died rapidly in the absence of supportive cytokines, ADP-520 demonstrated enhanced autonomous persistence throughout 16 days of culture ( Figure 5A ). The TRuC⁺ population within ADP-520 was enriched from ~60% TRuC⁺ in the starting population to >80% by Day 16 of culture ( Figure 5B ), in contrast to the modest enrichment of C10 TRuC T cells, indicating that the mbIL-15 enhancement acts in cis, promoting TRuC⁺ persistence. While NT and C10 TRuC T cells expanded significantly in response to IL-2 or IL-15 cytokine stimulation, ADP-520 maintained consistent cell numbers throughout 16 days of culture ( Supplementary Figure 5A ). The ADP-520 TRuC⁺ population did not enrich in the presence of exogenous IL-2 and decreased to a small extent in response to IL-15 ( Supplementary Figure 5A ). Thus, ADP-520 TRuC T cells display increased cytokine-free persistence and exhibit diminished response to exogenous γ-chain cytokines acting through IL-2Rβ.

Next, we investigated the impact of mbIL-15 on antigen-dependent persistence. TRuC T cells were challenged with CD70^hi 786-O tumor cells at an E:T ratio of 5:1 and T-cell numbers were assessed over time. After single stimulation with CD70^hi 786-O tumor cells, C10 TRuC T cells showed initial expansion during tumor cell clearance at Day 4, followed by T-cell contraction to Day 16 of culture ( Figure 5C ). ADP-520 killed tumor cells and expanded to Day 12 before reaching stasis at Day 16 ( Figure 5C ). Accordingly, ADP-520 demonstrated drastically improved viability of T cells throughout the 16-day culture compared with C10 TRuC T cells ( Figure 5D ). Similar results were observed with CD70^mod ACHN tumor cells ( Supplementary Figure 5B ). These data indicate that mbIL-15 enhancement promotes ADP-520 TRuC T-cell persistence and antigen-dependent expansion.

We further explored the mechanism of action of the mbIL-15-mediated enhancement of T-cell expansion using pharmacological inhibitors. To confirm that initial ADP-520 expansion is regulated by primary antigen stimulation, we added the Src kinase and TCR signaling inhibitor dasatinib (34) to the ADP-520–786-O co-culture at various timepoints ( Figure 5E ). The addition of dasatinib at Day 0 or Day 1 after co-culture blocked initial T cell activation and reduced overall ADP-520 T-cell expansion in this assay. However, dasatinib added subsequent to tumor clearance (Day 4 or Day 6) had no impact on ADP-520 expansion ( Figure 5E ). To confirm that this long-term T-cell persistence is driven by mbIL-15, we targeted the IL-15 signaling pathway using JAK1/2 inhibitor ruxolitinib and JAK3 inhibitor tofacitinib. As expected, both inhibitors blocked ADP-520 T-cell expansion when added at Day 0, Day 4 or Day 6 ( Figure 5F ). These findings substantiate the role of mbIL-15 signaling in the long-term expansion of ADP-520 T cells following tumor clearance.

We next sought to investigate the persistence of ADP-520 during conditions of repeated stress. ADP-520 TRuC T cells were cultured with CD70^mod ACHN tumor cells at a 1:1 E:T ratio and rechallenged with additional target cells every 4 days for a total of 16 days, with T-cell expansion and exhaustion evaluated by flow cytometry. While the C10 TRuC T cells never significantly expanded after initial or repeated challenge with ACHN, the ADP-520 TRuC T cells continued to expand throughout all three rechallenges ( Figure 6A ). Despite C10 and ADP-520 TRuC T cells showing similar tumor clearance (~60%) at Day 4, C10 failed to clear ACHN tumor cells while ADP-520 demonstrated durable and enhanced clearance of ACHN tumor during subsequent rechallenges ( Figure 6B ). ADP-520 also maintained significantly improved viability (~85-90%) throughout all rechallenges, whereas C10 TRuC T-cell viability was significantly decreased on Day 8 ( Figure 6C ). Only the ADP-520 TRuC T cells showed enrichment of TRuC⁺ T cells throughout all ACHN rechallenges ( Figure 6D ). Similar results were observed with CD70^high 786-O cells ( Supplementary Figure 5C ), as well MJ and HuT78 lymphoma lines ( Supplementary Figures 5D, E ). As expected, ADP-520 did not expand during re-challenge with CD70 KO target cell lines ( Supplementary Figure 5F ).

To further explore the mechanism of improved response and viability to repeated stress by ADP-520 TRuC T cells, we evaluated the expression of exhaustion markers on ADP-520 compared with C10 during repeated challenges with ACHN. Severe T-cell exhaustion is defined by the co-expression of several exhaustion markers, including PD-1, lymphocyte activation gene 3 (LAG-3) and T-cell immunoreceptor with Ig and ITIM domains (TIGIT). Both the C10 and ADP-520 TRuC T cells showed upregulated PD-1 expression equivalently at Day 4 in both CD4⁺ ( Figure 6E ) and CD8⁺ ( Figure 6F ) TRuC T cells. However, while C10 TRuC T cells maintained high PD-1 expression (~95% in CD4⁺, ~50% in CD8⁺) throughout the rechallenges, PD-1 expression was lower in ADP-520 by Days 8–16 ( Figures 6E, F ), an effect that was also observed in response to 786-O rechallenge ( Supplementary Figure 5G ). We found that ADP-520 showed a reduced frequency of PD-1⁺LAG3⁺TIGIT⁺ co-expressing CD4⁺ and CD8⁺ T cells at Day 4 after co-culture with ACHN ( Figures 6G, H ) and 786-O ( Supplementary Figure 5G ), suggesting that ADP-520 shows slower kinetics of exhaustion marker upregulation. This reduced PD-1⁺LAG3⁺TIGIT⁺ phenotype was maintained after subsequent re-challenges with ACHN tumor cells ( Figures 6G, H ). The distribution of T_CM, T_Naïve, T_EM, and T_EMRA populations was also evaluated during tumor rechallenge, however both single and chronic antigen stimulation let to rapid and near-complete differentiation of both C10 and ADP-520 TRuC T cells toward the T_EM subset (data not shown). This convergence limited our ability to resolve meaningful differences in memory phenotype associated with the long-term persistence benefit of mbIL-15. Overall, ADP-520 demonstrates improved target-dependent potency, survival and resistance to exhaustion during repeated challenge with CD70-expressing target cell lines.

To evaluate the in vivo antitumor efficacy of ADP-520, we established human RCC tumor xenograft models for CD70^hi (786-O) or CD70^mod (ACHN) RCC tumors in NSG-major histocompatibility complex I/II (MHC I/II) double knock-out (DKO) mice. The C10 TRuC T cells demonstrated complete clearance of 786-O tumors ( Figure 7A ) and suppression of ACHN tumor growth ( Figure 7B ). Conversely, ADP-520-treated tumors exhibited apparent growth progression upon T-cell infusion, followed by tumor clearance in the 786-O ( Figure 7A ) and ACHN models ( Figure 7B ), when evaluated by tumor caliper measurements. Bioluminescence measurements of tumor status demonstrated that ADP-520 clearly and potently drove enhanced 786-O ( Figure 7A ) and ACHN ( Figures 7B, C ) tumor clearance, suggesting that early increases in tumor volume reflect tumor pseudoprogression, wherein tumor infiltration by immune cells presents as artificial and transient tumor enlargement (14, 35–37). To explore this further, tumors and blood were collected from treated mice and evaluated by flow cytometry. Although the frequency of human CD45⁺ (hCD45⁺) T cells was similar in the C10- and ADP-520-treated 786-O tumors, the ADP-520-treated tumors had ~25-fold increased total T-cell numbers compared with the C10-treated tumors ( Figure 7D ). Relatively few T cells were found in the blood of mice treated with either C10 or ADP-520 ( Figure 7D ). Similarly, the ADP-520-treated ACHN tumors contained a significantly increased frequency and number of hCD45⁺ T cells, with comparable numbers of T cells observed in the blood ( Figure 7E ). ADP-520 similarly enhanced antitumor efficacy and intratumoral T-cell infiltration in the CD70^low NSCLC tumor model NCI-H1975 ( Supplementary Figures 6A, B ). These data demonstrate that ADP-520 TRuC T cells provide potent in vivo efficacy against a range of CD70 tumor models enabled in part by enhanced tumor infiltration.