IL-21 boosts T cell-therapy efficacy for solid tumors by enhancing mitochondrial Ca²⁺-mediated motility of effector CD8⁺ T cells

Abstract

Cell motility, characterized by random walk and exploratory search movement, enables effector CD8⁺ T cells to search for sparse antigen-specific cancer targets within a tumor. This is of special relevance for treatment of solid cancers with adoptive T-cell receptor (TCR) T-cell therapy, where administered effector CD8⁺ T cells recognize specific MHC-I–presented antigens. Cell motility requires cytoskeleton remodeling to facilitate shape changes and movement. Herein, we show that increased mitochondrial Ca²⁺ levels are essential to reduce cytoskeleton stiffness of effector CD8⁺ T cells, leading to acquisition of a polarized shape and high motility. IL-21, but not IL-7 or IL-15, was able to raise mitochondrial Ca²⁺ levels in effector CD8⁺ T cells and increase their motility without affecting survival and proliferation. This increase in mitochondrial Ca²⁺ levels triggered by IL-21 was driven by sustaining mitochondrial membrane potential through mitochondrial STAT3, independently of its transcriptional activity. Enhanced motility of effector CD8⁺ T cells led to a superior killing efficacy of antigen-specific melanoma cells in vitro. Furthermore, enhanced mitochondrial Ca²⁺-mediated motility of adoptive TCR-specific effector CD8⁺ T cells resulted in a superior antitumor efficacy of this treatment against solid tumors in vivo. Thus, enhancing effector CD8⁺ T-cell motility is a promising strategy to boost efficacy of adoptive T-cell therapies against solid tumors.

INTRODUCTION

Adoptive T-cell therapies have proven efficacious for the treatment of hematological malignancies (1-3), but clinical benefit remains limited in solid tumors. There are only two adoptive T-cell therapies approved for solid tumors: a tumor-infiltrating lymphocyte (TIL) therapy for unresectable or metastatic melanoma (4) and a T-cell receptor (TCR) T-cell therapy for unresectable or metastatic synovial sarcoma (5). Unlike hematological cancers, penetrance of T cells into solid tumors poses a unique obstacle to an effective T cell–mediated antitumor response (6). More importantly, after a few administered T cells manage to enter a solid tumor, they need to randomly move and search within the tumor microenvironment to find their target antigen–expressing cancer cells (7, 8). A highly motile CD8⁺ T cell has a greater capacity for random, exploratory movement and can interact with and kill more cancer cells than a less motile T cell, within the same timeframe.

While trafficking of T cells to tumors in the context of T-cell therapies has been extensively investigated (9, 10), little is known about intratumoral T-cell motility or which factors impact T-cell motility and can be leveraged to improve success of adoptive T-cell therapy. Cell motility is a tightly regulated, complex biological process dependent on coordinated cytoskeletal dynamics. Non-motile cells contain a rigid cytoskeleton characterized by crosslinked actin fibers and actin bundles, restricting their movement (11). In contrast, motile cells contain fewer cytoplasmic actin fibers and actin bundles, resulting in reduced cytoskeletal stiffness (12, 13). Mitochondria play a critical role in providing the localized Ca²⁺ required for dynamic actin cytoskeletal changes necessary for cell motility (14). Mitochondria serve as a major Ca²⁺ buffering organelle, maintain cytosolic Ca²⁺ homeostasis, and are in frequent contact with the endoplasmic reticulum (15). Mitochondrial Ca²⁺ has been found to facilitate cell migration through F-actin and focal adhesion dynamics, and impaired mitochondrial Ca²⁺ reduces cell migration by increasing cytoskeletal stiffness (14, 16, 17). Mitochondrial Ca²⁺ uptake is promoted by mitochondrial membrane potential (MMP) (18, 19). Unlike endoplasmic reticulum Ca²⁺, little is known about mitochondrial Ca²⁺ in T cells. In CD4⁺ T cells, mitochondrial Ca²⁺ is dispensable for proliferation and differentiation, but it increases the motility of CD4⁺ T cells activated in the presence of IL-6 (19-21). In CD8⁺ T cells, mitochondrial Ca²⁺ is also dispensable for proliferation, differentiation, and cytokine production (20). However, no studies have investigated the role of mitochondrial Ca²⁺ in CD8⁺ T-cell motility.

During the manufacturing of T-cell therapies, cells are expanded in the presence of cytokines to sustain proliferation. The most commonly used cytokine for ex vivo expansion of commercial T-cell therapies is IL-2 (22). However, there is growing interest in improving T-cell phenotype using additional cytokines. For example, T cells activated in the presence of IL-2 and IL-7 are more resistant to cell death compared to cells activated in IL-2 alone (23). Additionally, compared to T-cell expansion with IL-2, expansion with IL-7 and IL-15 has been shown to maintain stem-like memory T cells (24, 25). More recently, IL-21 is also being considered as it has been found to maintain a less-differentiated T-cell phenotype when added during activation or expansion with IL-2 (26, 27). However, little is known about the impact of these cytokines on CD8⁼ T-cell motility when added during cell expansion.

Here we show that mitochondrial Ca²⁺ is essential to sustain motility of effector CD8⁺ T cells by reducing cytoskeleton stiffness. IL-21, but not IL-7 or IL-15, was able to raise mitochondrial Ca²⁺ levels in effector CD8⁺ T cells through mitochondrial STAT3. Furthermore, mitochondrial Ca²⁺–driven motility of effector CD8⁺ T cells resulted in superior killing efficacy against melanoma cells in vitro and in vivo.

MATERIALS AND METHODS

Mice

Wild-type (WT) C57BL/6 J mice (RRID:_IMSR_JAX:000664), OT-I transgenic mice (RRID:IMSR_JAX:003831) and CD4-Cre mice (RRID:_IMSR_JAX:022071) were purchased from Jackson Laboratories. T-cell conditional MCU knockout (TcMCU KO) mice were previously described (21) and were generated by crossing homozygous floxed MCU mice (Mcu^flox/^flox) (28) with CD4-Cre transgenic mice. STAT3-S⁷²⁷A knockin-in (KI) mice were previously described (29). T-cMCU KO mice were crossed with OT-I transgenic mice to generate T-cMCU KO OT-I mice. Both male and female mice were used for these studies. Animal studies were approved by the University of Colorado Institutional Animal Care and Use Committee (IACUC) and performed in accordance with their guidelines.

Isolation, activation, and expansion of CD8⁺ T cells

Murine CD8⁺ T cells were isolated from spleen and lymph nodes of mice using negative selection as previously described (30), and were activated in the presence of Bruff’s medium [complete Click’s (EHAA) medium (cat# 05572, Sigma-Aldrich Inc., Saint Louis, MO) with β-mercaptoethanol (cat# M6250, Sigma-Aldrich Inc.), L-glutamine (cat# 25030081, Thermo Fisher Scientific, Waltham, MA), and 5% FBS (cat# PS-FB3, PEAK Serum, Bradenton, FL)] with plate-bound anti-CD3 (5 μg/ml, BioXcell, Lebanon, NH) and soluble anti-CD28 (1 μg/ml, BioXcell) for 48 hours at 10⁶ cells/ml. After 48 hours, CD8⁺ T cells were extensively washed, and expanded at 0.4 x 10⁶ cells/ml in Bruff’s medium with recombinant human IL-2 (rhIL-2; 40 IU/ml, R&D Systems, Minneapolis, MN) alone, or rhIL-2 in combination with recombinant mouse IL-21 (rmIL-21; 100 ng/ml, R&D Systems), rmIL-7 (10 ng/ml, R&D Systems), or rmIL-15 (100 ng/ml, R&D Systems). After 2 days (1st expansion), cells were split 1:3 using fresh Bruff’s medium and cytokine additions were made for another 2 days (2nd expansion).

For re-stimulation of murine effector CD8⁺ T cells, following two expansions in the presence of IL-2 or IL-2 plus IL-21, CD8⁺ T cells were washed, resuspended, and activated with plate-coated anti-CD3 (5 μg/ml, BioXcell) and soluble anti-CD28 (1 μg/ml, BioXcell) at 37°C. Supernatant was collected after 24 or 48 hours.

De-identified human blood was obtained from leukoreduction system (LRS) chambers (discard material) at the Colorado Children’s Hospital Blood Bank. PBMC were isolated from LRS chambers using histopaque (cat# 10771, Sigma-Aldrich Inc.). De-identified PBMC are considered no human subjects, no consent needed according to ethical standards. Human CD8⁺ T cells were isolated from these PBMC using positive selection CD8 Microbeads, human (cat# 130-045-201, Miltenyi Biotec, Gaithersburg, MD) and activated with plate-bound anti-CD3 at 10 μg/ml (cat# 300438, BioLegend, San Diego, CA, RRID:AB_11146991) and soluble anti-CD28 at 5 μg/ml (cat# 302933, BioLegend, RRID:AB_11150591) at 10⁶ cells/ml in RPMI (cat# 11875119, Life Technologies Corp., Carlsbad, CA) 10% FBS (cat# PS-FB3, PEAK Serum) for 48 hours as previously described (21). After 48 hours, human CD8⁺ T cells were extensively washed, and expanded at 0.5 x 10⁶ cells/ml in RPMI 10% FBS medium with rhIL-2 (100 IU/ml, R&D Systems) alone, or rhIL-2 in combination with rhIL-21 (100 ng/ml, R&D Systems). After 2 days (1st expansion), cells were split 1:2 or 1:3 using fresh RPMI 10% FBS medium and cytokine additions were made for another 2 days (2nd expansion).

Cell lines

B16 murine melanoma cell line was kindly provided by Dr. Ywen Zhu (2024, University of Colorado). B16-OVA murine melanoma cell line expressing ovalbumin (OVA) was kindly provided by Dr. Richard W. Dutton (2017, University of Massachusetts). B16 and B16-OVA cell lines were thawed and passaged in Bruff’s medium. B78mChOVA murine melanoma cell line lacking melanin and expressing OVA and mCherry was kindly provided by Dr. Matthew Krummel (2011, University of California San Francisco). B78mChOVA cell line was thawed and passaged in DMEM (cat# MT10027CV, Corning Inc., Kennebunk, ME) supplemented with β-mercaptoethanol (cat# M6250, Sigma-Aldrich Inc.), L-glutamine (cat# 25030081, Thermo Fisher Scientific), 10% FBS (cat# PS-FB3, PEAK Serum), and HEPES buffer (cat# 15630-080, Gibco, Waltham, MA). All cell lines were passaged every 2 days at a density of 1:4 and not used in experiments past four passages. Cell morphology and growth characteristics were monitored during culture and remained unchanged. The cell lines described above tested negative for Mycoplasma (Lonza mycoplasma detection kit). B16-OVA and B78mChOVA were authenticated by the response of OT-I CD8⁺ T cells to their OVA expression.

Live-cell confocal microscopy

For live confocal microscopy, 35mm Glass Bottom Dishes (MakTek Corp., Ashland, MA) were coated with 200 μl of rICAM-1-Fc Chimera (BioLegend) at 2 μg/ml in PBS, incubated overnight at 4°C, then washed with PBS. After the second expansion, CD8⁺T cells were washed, counted with trypan blue, diluted with Bruffs medium to 0.5-1 x 10⁶ cells/ml, and 150 μl of diluted cells were added to plates. Cell motility was recorded by time-lapse imaging of live cells with an inverted Zeiss LSM-780 scanning confocal microscope. Samples were mounted on the microscope in a PeCon PM2000RBT stage-top incubator and kept at 37°C. Using a 20X, 0.8NA Zeiss Plan-Apochromat air objective, the samples were scanned with the 561nm laser in brightfield or DIC mode. Sequential diffraction-limited images (1024 x 1024 pixels, 0.208 μm pixel size) were captured every 2 seconds for 10-20 minutes. Cell trajectory, speed, and distance traveled were determined using MtrackJ ImageJ plug-in (NIH, RRID:SCR_003070).

Transwell migration assay

For transwell migration, 500 μL of Bruffs medium was added to each well of 24-well tissue culture treated plates (Corning Inc.). 1-2 x 10⁵ CD8⁺ T cells were washed and placed into transwell inserts (6.5mm Insert, 5.0μm Polycarbonate Membrane: 3421, Corning Inc.) After 4-6 h, transwells were removed, and the viable cells that migrated through h the transwell were collected and counted by trypan blue staining. For inhibitor experiments, CD8⁺ T cells expanded in the presence of IL-2 or IL-2/IL-21 were pre-treated with vehicle, CCCP (10 μM, cat# C2759, Sigma), a mitochondrial uncoupler that dissipates MMP, CGP-37157 (10 μM, cat# 75450-34-9, Tocris Bioscience, Ellisville, MI), an inhibitor of mitochondrial Ca²⁺ efflux into the cytoplasm through the mitochondrial Na⁺/Ca²⁺ exchanger, or STA-21 (10 μM, cat# 28882-53-3, Cayman Chemical, Ann Arbor, MI), an inhibitor of STAT3 transcriptional activity, prior to assay.

Two-photon microscopy

After the second expansion with IL-2 or IL-2 plus IL-21, CD8⁺ T cells were labeled with CellTrace^™ Violet (cat# C34557, Invitrogen, Waltham, MA) or CFSE (cat# C34554, Invitrogen), extensively washed, resuspended in PBS, and a 1:1 mixture of cells was transferred by tail vein injection into naïve WT mice for lymph node imaging experiments or B78-mChOVA tumor-bearing WT mice for tumor imaging experiments.

Lymph Node Imaging.

Between 18 and 24 h after transfer, mice were euthanized and their lymph nodes (LNs) harvested for imaging. Explanted LNs were adhered to a coverslip using Vetbond (3M, Maplewood, MN). The LNs were placed in a PH-1 flow chamber (Warner instruments, Hamden, CT) and perfused with RPMI without phenol red medium (Gibco, Waltham MA), which was saturated with 95% O₂ and 5% CO₂ and kept between 35-37°C. Two-photon imaging was performed using a Leica SP8 DIVE upright two-photon microscope with a SpectraPhysics InsightX3 dual line (tunable 680-1300nm and 1045nm) infrared (IR) laser with pre-chirp compensation, 4 tunable spectral non-descanned detectors, galvanometer confocal scanner and high-speed resonant confocal scanner. Time-lapse imaging was done by repeated imaging of XY planes of 512x512 pixels at 1.16 μm/pixel and Z-steps of 3 μm with XYZ stacks acquired every 30 sec for 20-30 min. The acquired images were analyzed using Imaris software (Oxford Instruments, Abingdon, United Kingdom, RRID:SCR_007370). To eliminate any potential bleed-over between channels the images were linearly unmixed using the Channel Arithmetic function in Imaris to subtract any non-specific fluorescence bleeding from one channel to the other as previously described (31). Cells in LNs that could be tracked for ≥5 min were included in the analysis, and mean square displacement, average speed, and total displacement were calculated.

Tumor Imaging.

B78mChOVA cells were extensively washed in PBS, counted by trypan blue staining, and 0.6 x 10⁶ cells in 150 μL PBS were injected subcutaneously into WT naïve mice. Five days post implantation, OT-I cells were adoptively transferred by tail vein. Two days post OT-I adoptive transfer, mice were euthanized and tumors harvested for imaging. Explanted tumors were adhered to a coverslip using Vetbond (3M). The tumors were placed in a flow chamber and perfused as described above. Two-photon imaging was performed as described above, with XYZ stacks acquired every 30 sec for 10-30 min. The acquired images were analyzed using Imaris software (Oxford Instruments), and OT-I cells in tumors that could be tracked for ≥3 min were included in the analysis and mean square displacement, average speed, and total displacement were calculated.

Flow cytometry

For analysis of cell surface markers, cells were stained with fluorescence-conjugated antibodies: anti-CD4 AF647 (BioLegend Cat# 100530, RRID:AB_389325), anti-CD8 PacBlue (BioLegend Cat# 100725, RRID:AB_493425), anti-CD11b PE (BioLegend Cat# 101208, RRID:AB_312791), anti-CD44 BV510 (BioLegend Cat# 103043, RRID:AB_2561391), anti-CD62L PEDazzle594 (BioLegend Cat# 104447, RRID:AB_2566162), anti-CD25 FITC (BioLegend Cat# 102006, RRID:AB_312855), anti-CD69 PerCP (BioLegend Cat# 104520, RRID:AB_940495), anti-CCR7 PerCP/Cy 5.5 (BioLegend Cat# 120115, RRID:AB_2228801), and anti-CD127 BV510 (BioLegend Cat# 135033, RRID:AB_2564576). Cells were examined by flow cytometry analysis using Northern Lights and Aurora, Cytek Biosciences flow cytometers. Flow cytometry analyses were performed by FlowJo v.10.10.0 (RRID:SCR_008520). The live population was gated based on FSC-A and SSC-A and singlets based on FSC-A and FSC-H.

Cytokine production by CD8⁺ T cells was examined by intracellular staining and flow cytometry. Cells were fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences, Milpitas, CA). Cells were then stained following the recommended procedure by the manufacturer with anti-IFNγ AF647 (BD Biosciences Cat# 557735, RRID:AB_396843), anti-IL-2 PE (BioLegend Cat# 503808, RRID:AB_315302), anti-TNFa AF647 (BioLegend Cat# 506314, RRID:AB_493330), anti-T-bet BV605 (BioLegend Cat# 644817, RRID:AB_11219388), and anti-Perforin APC (Thermo Fisher Scientific,, Cat# 17-9392-80, RRID:AB_469514). Cells were washed and examined on the flow cytometer as described above (Northern Lights and Aurora, Cytek Biosciences).

For MMP analysis, cells were stained with TMRE (Invitrogen) at 0.5 μM for 20 minutes at 37°C and examined by flow cytometry analysis, as previously described (32). For mitochondrial Ca²⁺ analysis, cells were stained with Rhod-2 AM (Invitrogen) at 5 μM for 30 minutes at 37°C and examined by flow cytometry analysis, as previously described (21). For mitochondria mass analysis, cells were stained with Mitotracker Deep Red 633 (Invitrogen, 1:1500 dilution) for 30 minutes at 37°C, and examined by flow cytometry analysis, as previously described (33). For inhibitor experiments, CD8⁺ T cells expanded in the presence of IL-2 or IL-2/IL-21 were pre-treated with vehicle, CCCP (10 μM, cat# C2759, Sigma), or STA-21 (10 μM, cat# 28882-53-3, Cayman Chemical), prior to analyses of MMP and mitochondrial Ca²⁺.

For CD107a degranulation analysis, CD8⁺ T cells were incubated with fluorescent anti-CD107a AF700 (BioLegend Cat# 121628, RRID:AB_2783063) at 37°C during the last 4 hours of restimulation. Cells were washed and CD107a expression was examined by flow cytometry as described above.

Cytosolic Ca²⁺ measurement

Cytosolic Ca²⁺ was measured by staining cells with Fura-2 AM (Thermo Fisher) at 5 μM for 30 min, followed by fluorometric measurement (340/380 excitation) in a Tecan microplate reader as previously described (19). F340/380 ratio was calculated by dividing the fluorescence reading at 340 nm by the fluorescence at 380 nm as previously described (19).

Confocal microscopy

CD8⁺ T cells were cytospun on slides, fixed with 3.7% formaldehyde, and stained with rhodamine phalloidin (Invitrogen) at 1:400 dilution (F-actin) and DAPI (nuclear marker), as previously described (21). Slides were imaged on an inverted Zeiss LSM780 scanning confocal microscope using a 100x, 1.4NA Zeiss Plan-Apochromat objective. Contour lines along the cytoplasmic membrane were drawn and area of rhodamine fluorescence beneath the cytoplasmic membrane was quantified using FIJI (RRID:SCR_002285). For Mitotracker analysis, cells were stained with Mitotracker Green (cat# M7514, Invitrogen) for 30 minutes at 37°C, cytospun on slides, fixed with 3.7% formaldehyde, and stained with DAPI (21). Slides were imaged on an inverted Zeiss LSM780 scanning confocal microscope, 1.4NA Zeiss Plan-Apochromat objective. For inhibitor experiments, CD8⁺ T cells expanded in the presence of IL-2 or IL-2/IL-21 were treated with vehicle, CCCP (10 μM, cat# C2759, Sigma), or CGP-37157 (10 μM, cat# 75450-34-9, Tocris Bioscience), prior to cytospin.

Atomic force microscopy

After the second expansion, CD8⁺ T cells were harvested and resuspended in XF RPMI Seahorse Medium (Agilent Technologies). TPP AFM dishes (Cat#93040) were coated with 22.2ng/ml Cell-Tak^™ (Corning) and incubated for 1 h prior to cell seeding. Plates were washed and CD8⁺ T cells were added at 0.3 x 10⁶ cells/ml and incubated for 2 h at 37 °C. Atomic Force Microscopy (AFM) measurements were performed using a JPK NanoWizard 3a system (software v6) integrated with a Zeiss AxioObserver microscope and equipped with Bruker MSNL-10 probes. Cantilever C, with a nominal spring constant of 0.01 N/m, was selected for all experiments. The cantilever was calibrated both on the substrate surface and under thermal oscillation using a 2 V setpoint and a 3 μm z-length to ensure consistency with the manufacturer-specified spring constant. AFM measurements were conducted at room temperature to minimize cell motility. For each cell, five force–distance curves were collected using a 1 nN indentation force and a maximum indentation depth of 5 μm. Young’s modulus was calculated from each curve using the Hertz–Sneddon model with a triangular pyramidal tip geometry in the JPK Data Processing software (v6.0.77).

Western blot analysis

Whole cell extracts were prepared in RIPA lysis buffer supplemented with 1 mM PMSF, 1 mM Na₃VO₄, and 0.5% Protease Inhibitor Cocktail (Sigma Aldrich Inc.). Mitochondrial extracts were prepared using the Mitochondrial Fractionation Kit (Active Motif, Carlsbad, CA) according to the manufacturer’s recommendations. Briefly, cells were lysed while preserving mitochondrial integrity. Lysates were subjected to sequential centrifugation steps to remove debris, followed by high-speed centrifugation to pellet mitochondria. The resulting supernatant was collected as the cytosolic fraction, while the mitochondrial pellet was washed and resuspended in mitochondrial buffer. Protein concentration of each lysate was determined by Bradford Assay (Bio-Rad, Hercules, CA). Western blot analyses on whole cell extracts and mitochondrial extracts was performed as previously described (21). Antibodies used were as follows: anti-MCU (cat#14997 Cell Signaling Technologies (CST), Danvers, MA, RRID:AB_2721812), anti-Total STAT3 (CST cat#9139S, RRID:AB_331757), anti-pSTAT3 (Y705) (CST cat#9145S, RRID:AB_2491009), anti-pSTAT3 (S727) (CST cat#9134S, RRID:AB_331589), anti-GAPDH (CST cat#2118S, RRID:AB_561053), anti-β-Actin (CST cat#4967S, RRID:AB_330288), anti-α-tubulin (CST cat#2144S, RRID:AB_2210548), anti-UQCR2 (Abcam, Waltham, MA, Ab14745, RRID:AB_2213640), and anti-NDUFA9 (Abcam Ab14713, RRID:AB_301431). Secondary antibodies, anti-rabbit IgG and anti-mouse IgG (Jackson Immunologicals, West Grove, PA), were added at 1:2500 dilution. Blots were developed using KPL LumiGLO chemiluminescent substrate system (SeraCare Life Sciences, Milford, MA) with Genesys G:Box Chemi-XX6.

Cytokine production analysis

Supernatant from the CD8⁺ T-cell culture at different time points was collected and stored at -80°C until used to quantify cytokine levels. Murine IFNγ and IL-2 concentrations in the sample supernatant were determined using a sandwich ELISA as previously described (33) with anti-mouse IFNγ capture and biotinylated antibodies (BioLegend Cat# 505702, RRID:AB_315390 and Cat# 505804, RRID:AB_315398) or anti-mouse IL-2 capture and biotinylated antibodies (BioLegend Cat# 503702, RRID:AB_315292 and Cat# 503804, RRID:AB_315298). Granzyme A levels were quantified using a mouse Granzyme A ELISA kit (Abcam ab253213) according to the manufacturer’s instructions. All samples and standards were run in duplicate, and samples were loaded in biological triplicate. Absorbance was measured using a microplate reader at 450nm. Cytokine concentrations were calculated in relation to the standard curve.

In vitro cancer cell killing assay

B16-OVA and B16 (non-OVA) cells were plated in a 24-well plate and grown for 48 hours before adding IL-2– or IL-2/IL-21–expanded OT-I or T-cMCU KO OT-I CD8⁺ T cells for co-culture at the indicated effector-to-target (E:T) ratios (1:1, 0.5:1). Prior to OT-I cell addition, the number of B16 cells was determined by trypan blue staining. For inhibitor experiments, OT-I CD8⁺ T cells expanded in the presence of IL-2 or IL-2/IL-21 were treated with vehicle orCGP (10 μM, cat# 75450-34-9, Tocris Bioscience), which inhibits mitochondrial Ca²⁺ efflux into the cytoplasm through the mitochondrial Na⁺/Ca²⁺ exchanger, prior to co-culture for 7 h at 37°C. The cells were then washed and added to the B16-OVA/B16 cells for co-culture. The cancer cells were co-cultured with the OT-I T cells for 24 h at 37°C. Following co-culture, the number of live B16-OVA cells was quantified by trypan blue staining. T-cell killing was calculated as the total number of live cancer cells in co-culture wells relative to the number of live cancer cells cultured for the same time in the absence of OT-I cells.

For the experiments using non-OT-I CD8⁺ T cells (WT and STAT3 KI CD8⁺ T cells), total spleen cells (10⁶ cells/ml in 96 well plate) were activated with N4 OVA peptide (5 μM, OVA 257-264, Cat: vac-sin, InvivoGen, San Diego, CA) and IL-2 (25 U/ml) for 5 days. Activated CD8⁺ T cells were then expanded with IL-2 or IL-2 plus IL-21 and added to B16-OVA melanoma cells at 4:1 E:T ratio. The number of live B16-OVA cells was quantified 24 h later and T-cell killing was calculated as described above.

In vivo melanoma studies

A 1:1 mixture of B16-OVA:B16 cells (4x10⁵ cells/mouse) was implanted subcutaneously into the right flank of WT mice. OT-I CD8⁺ T cells or T-cMCU KO OT-I CD8⁺ T cells were isolated, activated, and expanded with IL-2 or IL-2/IL-21 as described above, and 5x10⁵ cells/mouse were administered intravenously to B16/B16-OVA–tumor bearing mice 8 days post tumor implantation. Mice were monitored every other day during the first week and daily thereafter. Tumor size was followed over time and measured with calipers. Tumor volume was calculated as Vol = length x width x height. As humane endpoints, mice were euthanized when the tumor reached 2 cm in any one dimension, tumors were ulcerated or infected, or if any major sign of discomfort presented. Following the adoptive transfer of the OT-I CD8⁺ T cells, tumor growth was measured over a period of time.

For flow cytometry analysis of intratumoral OT-I CD8⁺ T cells, 400,000 B16-OVA cells were subcutaneously implanted into the right flank of WT mice. OT-I or TcMCU KO OT-I CD8⁺ T cells were isolated, activated, and expanded with IL-2 or IL-2/IL-21 as described above. After the second expansion with IL-2 or IL-2 plus IL-21, OT-I or TcMCU KO OT-I CD8⁺ T cells were labeled with CellTrace^™ Violet or CFSE, extensively washed, resuspended in PBS and a 1:1 mixture of 1 x 10⁶ total OT-I cells was transferred by tail vein injection into B16-OVA–tumor bearing mice 10 days post tumor implantation. Two days post-OT-I adoptive transfer, mice were euthanized, and tumors were processed according to the Miltenyi Tumor Dissociation protocol (cat 130-096-730, Miltenyi Biotec) and digested using a GentleMACS Octo Dissociator, per manufacturer’s recommendations. Tumor samples were stained with anti-CD45 PerCP (BioLegend Cat# 103129, RRID:AB_893343) and anti-CD8 PacBlue (BioLegend Cat# 100725, RRID:AB_493425) and percentage of adoptively transferred cells was detected based on CFSE and CTV fluorescence by flow cytometry (as described above, see Flow cytometry).

Statistical analysis

Statistical analyses were performed using GraphPad Prism v.10 (RRID:SCR_002798). The standard Unpaired t-test was performed between two groups. The One-way ANOVA analysis was performed to compare means across three or more groups based on one independent variable. The Two-way ANOVA analysis was performed for more than two groups when analyzing the effect of two independent variables. Kaplan–Meier survival curves were analyzed with a log-rank (Mantel-Cox) test. Tumor growth curves were analyzed with Mixed-effects analysis. The standard p < 0.05 served as the cut-off for significance.

DATA AVAILABILITY

All data generated in this study are included in the article or uploaded as Supplementary Data. The raw data are also available on request from the corresponding author.

RESULTS

IL-21 enhances the motility of effector CD8⁺ T cells during their expansion with IL-2

Motility of effector CD8⁺ T cells (i.e. random walk and exploratory spread) remains poorly investigated despite being essential for these cells to search for and reach their cell targets. This is highly relevant in the context of cancer T-cell therapies, where effector CD8⁺ T cells are extensively grown in vitro with IL-2, maintaining their rounded morphology to facilitate cell division while losing their motility. Since the addition of other cytokines, such as IL-7, IL-15, and IL-21, during the expansion of effector T cells with IL-2 is now an emerging field to improve adoptive T-cell therapies (34), we investigated the impact of these cytokines on effector CD8⁺ T-cell polarization. CD8⁺ T cells isolated from LNs and spleen were activated with anti-CD3 and anti-CD28 (anti-CD3/CD28). After two days of activation, CD8⁺ T cells were harvested, washed, and expanded in the presence of IL-2 alone or IL-2 in combination with either IL-7, IL-15, or IL-21. After 2 days (1^st expansion), cells were split and expanded for another two days (2^nd expansion) in the presence of the corresponding cytokines (Fig. 1A). Visual examination (bright field microscopy) of the cells in culture revealed a higher frequency of polarized (i.e. elongated) CD8⁺ T cells in the presence of IL-2 plus IL-21 relative to mostly rounded cells in the presence of IL-2 or IL-2 plus IL-7 or IL-2 plus IL-15 (Supplementary Fig. S1A and S1B). To verify these observations, CD8⁺ T cells expanded under the different conditions were cytospun, stained with rhodamine phalloidin to visualize cellular morphology, and examined by confocal microscopy under low magnification for quantification. Effector CD8⁺ T cells expanded with IL-2 alone exhibited a predominantly round morphology, with very few polarized cells (characterized by formation of a leading edge and uropod), as expected (Fig. 1B and representative images in Supplementary Fig. S2). The addition of IL-7 or IL-15 during expansion with IL-2 did not significantly increase the frequency of polarized CD8⁺ T cells relative to IL-2 alone (Fig. 1B and Supplementary Fig. S2) and had no effect on the proliferation of the cells (Supplementary Fig. S3). However, a high frequency of CD8⁺ T cells expanded with IL-2 plus IL-21 were polarized cells (Fig. 1B and Supplementary Fig. S2).

Figure 1. The presence of IL-21 during the expansion of effector CD8⁺ T cells with IL-2 selectively increases their motility.

(A) Experimental design scheme. CD8 cells were isolated from WT C57BL/6 mice and activated with anti-CD3 (5 μg/ml) and anti-CD28 (1 μg/ml) Abs. After 2 days, cells were harvested, washed and cultured in the presence of IL-2 (40 U/ml) alone, IL-2 (40 U/ml) plus IL-21 (100 ng/ml), IL-2 (40 U/ml) plus IL-7 (10 ng/ml), or IL-2 (40 U/ml) plus IL-15 (100 ng/ml). Two days later (1st expansion), cells were split, provided with additional cytokines, and expanded for another 2 days (2nd expansion). (B) After the 2nd expansion, cells were cytospun, stained with rhodamine phalloidin and DAPI and imaged with a scanning confocal microscope (20X magnification). Subsequently, the frequency of polarized cells per image was examined, n= 4-5 number of images per condition. (C-E) CD8 cells were activated and expanded as in (A) in the presence of IL-2 alone or IL-2 plus IL-21 for two expansions (C). Number of viable CD8 cells was determined by trypan blue staining, n=3 (D). After the 2nd expansion, CD8 cells (1.0 x 10⁵) were set in transwells and the number of cells that migrated to the well were counted 4 h later, n=3 (E). (F-H) CD8 cells were activated and expanded as in (C). After the 2nd expansion, live-cell time-lapse microscopy was performed over 10 mins (20X magnification). Individual cell trajectory, normalized to coordinate axis (F), total distance traveled per cell (G), and average speed per cell (H) analyzed by FIJI MtrackJ. n= 68 (IL-2) or 47 (IL-2/IL-21) number of cells. (I-K) CD8 cells were activated and expanded as in (C). Expanded CD8 cells were labeled with CFSE (IL-2 alone) or Cell Trace Violet (IL-2 plus IL-21), and co-injected i.v. 1:1 into WT mice. Lymph nodes were harvested 24 h later and cell motility was measured using two-photon microscopy. Representative image showing trajectories for CD8 cells expanded with IL-2 (green) or IL-2/IL-21 (blue) (I), mean square displacement (J), total displacement (K), and average speed (L) per cell measured with Imaris software, n=152 (IL-2) and 47 (IL-2/IL-21) number of cells over 4 videos. * denotes p < 0.05 and ns denotes not significant as determined by One-Way ANOVA with multiple comparisons (B), Unpaired t-test (D-E, G-H, K-L), or Two-Way ANOVA (J). Mean ±SD is shown (B, D, E, G, H). Mean ±SEM is shown (J). Maximum, minimum, median, 25^th percentile and 75^th percentile are shown (K, L). All panels are representative of 2-5 independent experiments.

Despite its effect on effector CD8⁺ T-cell polarity, the presence of IL-21 during expansion with IL-2 (Fig. 1C) had no effect on the proliferation of CD8⁺ T cells relative to the proliferation with only IL-2 during the first expansion (Supplementary Fig. S4) or even during the second expansion (Fig. 1D). Effector CD8⁺ T cells did not proliferate when we tried to expand them with IL-21 alone (Supplementary Fig. S4), which is consistent with the fact that IL-21 is not a growth factor for T cells (35). We also examined whether the presence of IL-21 affected the phenotype of IL-2–expanded CD8⁺ T cells. There was no difference in the phenotype between effector CD8⁺ T cells expanded with IL-2 plus IL-21 relative to those expanded with IL-2 alone, as determined by the expression of CD44, CD69, and CD25 (Supplementary Fig. S5A). In addition, there was no difference in the production of IFNγ (Supplementary Fig. S5B), IL-2 (Supplementary Fig. S5C), TNFα (Supplementary Fig. S5D), and perforin (Supplementary Fig. S5E), or cell surface expression of CD107a (indicator of degranulation) (Supplementary Fig. S5F) following restimulation with anti-CD3/CD28 between effector CD8⁺ T cells expanded with IL-2 plus IL-21 and those expanded with IL-2 alone. Thus, the presence of IL-21 during the expansion of effector CD8⁺ T cells with IL-2 did not cause a major phenotypic change or TCR-mediated stimulation difference despite its effect on cell polarity.

We then examined whether the high frequency of polarized cells following the expansion of effector CD8⁺ T cells with IL-2 plus IL-21 translated to increased motility in a transwell motility assay. CD8⁺ T cells were activated, expanded with IL-2 alone or IL-2 plus IL-21, and following the second expansion, cells were washed and placed in transwells in medium alone for 4 h. The number of cells that migrated through the transwell to the lower chamber was counted. A higher number of IL-2/IL-21–expanded effector CD8⁺ T cells was able to travel through the transwell (Fig. 1E), indicative of greater motility for CD8⁺ T cells expanded with IL-2 plus IL-21 relative to IL-2–expanded CD8 cells. In contrast to IL-21, the presence of IL-7 or IL-15 during the expansion with IL-2 did not improve the motility of effector CD8⁺ T cells (Supplementary Fig. S6A). Since IL-15 and IL-21 have both been shown to promote memory phenotypes in CD8 cells (36, 37), we examined whether the selective effect of IL-21 on motility could be due to a potential different phenotype. However, we saw no difference between the phenotype of CD8⁺ T cells expanded with IL-2 plus IL-21 and those expanded with IL-2 plus IL-15, and both had more effector-like phenotype (high T-bet and CD44) than memory-like phenotype (low CCR7 and CD127), as expected for CD8⁺ T cells expanded following activation (Supplementary Fig. S6B).

To further demonstrate the enhanced motility of CD8⁺ T cells expanded with IL-2 plus IL-21, we used brightfield live-cell time-lapse microscopy. After the second expansion with IL-2 or IL-2 plus IL-21, effector CD8⁺ T cells were washed, resuspended in medium alone, and loaded onto ICAM-coated microscope plates for time-lapse live-cell microscopy. CD8⁺ T cells expanded with IL-2 plus IL-21 traveled more diverse trajectories (i.e. random walks) (Fig. 1F), longer total distance (Fig. 1G), and with higher average speed (Fig. 1H and Supplementary Videos S1 and S2) compared to those expanded with IL-2 alone. In addition, we quantified the number of polarized cells on the ICAM-coated plates from live-cell brightfield microscope images (Supplementary Fig. S7A). Effector CD8⁺ T cells expanded with IL-2 plus IL-21 exhibited a higher frequency of polarized cells (Supplementary Fig. S7B).

We also investigated whether the enhanced motility of effector CD8⁺ T cells expanded with IL-2 plus IL-21 is sustained in vivo when these cells were adoptively transferred into mice. Isolated CD8⁺ T cells were activated, expanded with IL-2 or IL-2 plus IL-21 for two expansions, fluorescently labeled with CFSE or CTV, respectively, and i.v. co-administered together at 1:1 ratio into WT mice. After 24 h, LNs were harvested, and the motility of the transferred cells within intact LNs was measured using time-lapse imaging with two-photon microscopy. Effector CD8⁺ T cells expanded with IL-2 plus IL-21 on average traveled longer paths (Fig. 1I), with greater mean square displacement (Fig. 1J), greater total displacement (Fig. 1K), and higher average speed (Fig. 1L and Supplementary Videos S3-S6), compared to those expanded in IL-2 alone.

In addition, to test whether the effect of IL-21 on motility could also be translated to human CD8⁺ T cells, we isolated CD8⁺ T cells from human healthy donor PBMC and activated them with anti-CD/CD28. As for mouse CD8⁺ T cells, 2 days after activation, human CD8⁺ T cells were harvested, washed, expanded in the presence of IL-2 alone or IL-2 plus IL-21 for two expansions (Supplementary Fig. S8A). We first examined the effect of IL-21 on cell polarity by staining with rhodamine phalloidin and visualizing the cells by confocal microscopy. The frequency of polarized cells was higher in human effector CD8⁺ T cells expanded with IL-2 plus IL-21 compared to IL-2 alone (Supplementary Fig. S8B and S8C). Moreover, analysis of cell motility by transwell assay revealed an increased motility of effector CD8⁺ T cells expanded with IL-2 plus IL-21 (Supplementary Fig. S8D).

Together, these results show that the presence of IL-21 during the expansion of mouse and human effector CD8⁺ T cells with IL-2 selectively enhanced their motility both in vitro and in vivo.

IL-21–induced effector CD8⁺ T-cell motility is mediated by increased mitochondrial Ca²⁺

The presence of high-density actin bundles in the cytoplasm has been shown to result in high cytoskeleton stiffness and reduced cell motility (14). We therefore examined the presence of cytoplasmic F-actin fibers in mouse effector CD8⁺ T cells expanded with IL-2 or IL-2 plus IL-21, by staining with rhodamine phalloidin and imaging with scanning confocal microscopy. As predicted based on their increased motility, there was less density of F-actin fibers (indicative of reduced cytoskeleton stiffness) within the cytosol of CD8⁺ T cells expanded with IL-2 plus IL-21 relative to IL-2–expanded CD8⁺ T cells (Fig. 2A).

Figure 2. IL-21 enhances the motility of effector CD8⁺ T cells by increasing mitochondrial Ca²⁺.

(A-B) WT CD8 cells were activated and expanded with IL-2 or IL-2/IL-21 as in Figure 1C. After the 2^nd expansion, cells were cytospun and stained with rhodamine phalloidin (red) and DAPI (blue) and imaged with a scanning confocal microscope (100X magnification). Representative images (left panel) and quantification (right panel) of cytoplasmic F-actin area/cell are shown. n=11 (IL-2) and 7 (IL-2/IL-21) number of cells (A). (B) CD8 cells were activated and expanded as in (A), and after two expansions cellular stiffness was determined by atomic force microscopy. Mean of Young's modulus is shown. n=15 (IL-2) and 12 (IL-2/IL-21) number of cells. (C-D) CD8 cells were activated and expanded as in (A), and after two expansions mitochondrial Ca²⁺ was determined by Rhod2-AM staining and flow cytometry. Representative histogram (C) and MFI quantification are shown, n=3 samples (D). (E-F) During the last 7 h of the 2^nd expansion, IL-2/IL-21-expanded CD8 cells were treated with vehicle or CGP-37157 (10 μM), washed, cytospun and stained with rhodamine phalloidin and DAPI. Quantification of cytoplasmic F-actin area/cell is shown. n=9 (IL-2), 12 (IL-2/IL-21) and 11 (IL-2/IL-21+CGP) number of cells (E). CD8 cells were expanded, treated with vehicle or CGP as in (E), washed and placed into transwells for motility assay. Number of cells migrated into the wells is shown, n=3 samples (F). (G-H) CD8 cells were isolated from WT and T-cMCU KO mice, activated with anti-CD3/CD28 Abs for 48 h, washed, and expanded with IL-2 or IL-2/IL-21. After the 2^nd expansion, IL-2/IL-21-expanded CD8 cells were seeded onto ICAM-Fc coated plates and the frequency of polarized cells was assessed in live-cell brightfield microscopy, n=8 images (G). After the 2^nd expansion, CD8 cells were placed into transwells for motility assay. Number of cells migrated into the wells is shown, n=3 samples (H). * denotes p < 0.05 as determined by Unpaired t-test (A, B, D, G), One-Way ANOVA with multiple comparisons (E), or Two-Way ANOVA with multiple comparisons (F, H). Mean ±SD is shown. All panels are representative of 2-3 independent experiments.

To further demonstrate the difference in cytoskeleton stiffness between effector CD8⁺ T cells expanded with IL-2 and those expanded with IL-2 plus IL-21, we performed AFM. AFM measures the elastic modulus (Young's Modulus) of a cell at the indentation depths of a few hundred nanometers after applying a predetermined force to an individual live cell through a cantilever tip. Higher Young’s Modulus is indicative of higher cytoskeleton stiffness. AFM analyses showed lower Young's Modulus in CD8⁺ T cells expanded with IL-2 plus IL-21 relative to the Young's Modulus for CD8⁺ T cells expanded with IL-2 only (Fig. 2B). Thus, the presence of IL-21 during the expansion of effector CD8⁺ T cells reduces their cytoskeletal stiffness.

One known factor influencing cytoskeletal stiffness and cell motility is mitochondrial Ca²⁺ levels (14). Defects in mitochondrial Ca²⁺ uptake and mitochondrial Ca²⁺ efflux have been shown to impair actin polymerization and depolymerization events as well as reduce cell motility (14, 21, 38). We therefore investigated whether the presence of IL-21 during effector CD8⁺ T-cell expansion with IL-2 has any impact on mitochondrial Ca²⁺ levels, using Rhod2-AM staining and flow cytometry analysis. CD8⁺ T cells expanded with IL-2 plus IL-21 had elevated mitochondrial Ca²⁺ levels compared to IL-2–expanded CD8⁺ T cells (Fig. 2C and 2D), indicating that IL-21 increased mitochondrial Ca²⁺ levels in effector CD8⁺ T cells. In contrast, neither IL-7 nor IL-15 had an effect on mitochondrial Ca²⁺ levels (Supplementary Fig. S9A). Moreover, mitochondrial Ca²⁺ levels were also increased in human CD8⁺ T cells expanded with IL-2 plus IL-21 (Supplementary Fig. S9B and S9C). Thus, the presence of IL-21 during the expansion of effector CD8⁺ T cells with IL-2 selectively increased mitochondrial Ca²⁺ levels.

As expected, considering that mitochondria normally generate spatially confined increases in the levels of Ca²⁺ within the cytosol, there was a small but significant increase in total cytosolic Ca²⁺ levels in mouse CD8⁺ T cells expanded with IL-2 plus IL-21 relative to the levels in IL-2–expanded CD8⁺ T cells (Supplementary Fig. S9D). To determine whether the decreased cytoskeleton stiffness in CD8⁺ T cells expanded with IL-2 plus IL-21 was facilitated by the increased levels of mitochondrial Ca²⁺ triggered by IL-21, we tested the effect of a pharmacological inhibitor of mitochondrial Ca²⁺ release into the cytosol. The uptake of Ca²⁺ into mitochondria (from ER or extracellular Ca²⁺) is mediated by the mitochondrial Ca²⁺ uniporter (MCU) protein complex (39). Being highly dynamic, mitochondria then deliver their Ca²⁺ to given locations (e.g. actin fibers) within the cytosol; the release of mitochondrial Ca²⁺ into the cytosol occurs through the mitochondrial Na⁺/Ca²⁺ exchanger (mNCX) (39). WT CD8⁺ T cells were isolated, activated, and expanded with IL-2 or IL-2 plus IL-21. During the last 7 h of the second expansion, IL-2/IL-21–expanded cells were treated with CGP-37157 (CGP), a pharmacological inhibitor of mNCX (40), cytospun, stained with rhodamine phalloidin to measure F-actin, and visualized by confocal microscopy. Treatment with CGP increased the presence of F-actin bundles in CD8⁺ T cells expanded with IL-2 plus IL-21 (Fig. 2E). Thus, IL-21–induced changes to the F-actin cytoskeleton by IL-21 were dependent on mitochondrial Ca²⁺.

To then test whether mitochondrial Ca²⁺ contributed to the increased motility of effector CD8⁺ T cells expanded in the presence of IL-21, cells treated with vehicle or CGP were washed and seeded for the transwell assay with medium alone. The enhanced motility of CD8⁺ T cells expanded with IL-2 plus IL-21 was abrogated by mitochondrial Ca²⁺ efflux inhibition (Fig. 2F). The effect of pre-treatment with CGP on the motility of IL-2/IL-21–expanded CD8⁺ T cells was not due to a toxic effect on cell viability, as determined by live/dead staining and flow cytometry analysis post-treatment (Supplementary Fig. S10A). In addition, we restimulated IL-2/IL-21–expanded effector CD8⁺ T cells treated with CGP or vehicle with anti-CD3/CD28, and examined their production of IL-2 and granzyme A. There was no impact of CGP pre-treatment on IL-2 or granzyme A production (Supplementary Fig. S10B and S10C). Thus, pharmacological inhibition of mitochondrial Ca²⁺ selectively abrogated the superior motility of IL-2/IL-21–expanded effector CD8⁺ T cells without affecting their survival or TCR-dependent effector function.

To further demonstrate the contribution of mitochondrial Ca²⁺ to IL-21-induced motility in effector CD8⁺ T cells, we used CD8⁺ T cells from T-cMCU KO mice obtained by crossing MCU^f/f with CD4-Cre transgenic mice. T-cell development in the thymus and the presence of CD4⁺ and CD8⁺ T cells in the spleen and LNs are not affected in T-cMCU KO mice (21). We examined whether the lack of MCU could impact the activation of CD8⁺ T cells. CD8⁺ T cells were isolated from WT and T-cMCU KO mice and activated with anti-CD3/CD28. Analysis of viable cell number after activation showed no difference between WT and T-cMCU KO CD8⁺ T cells (Supplementary Fig. S11A). Additionally, analyses of CD25, CD44, and CD69 surface activation markers by flow cytometry showed no difference between WT and T-cMCU KO CD8⁺ T cells (Supplementary Fig. S11B). We then examined the potential effect of MCU deficiency on effector CD8⁺ T cells following their expansion with IL-2 or IL-2 plus IL-21. We first verified the absence of MCU in expanded T-cMCU KO CD8⁺ T cells, by Western blot analysis of whole cell extracts (Supplementary Fig. S11C). No difference in cell proliferation was observed between WT and T-cMCU KO CD8⁺ T cells expanded with IL-2 or IL-2 plus IL-21 (Supplementary Fig. S11D). However, there was a marked reduction in the frequency of polarized T-cMCU KO CD8⁺ T cells expanded with IL-2 plus IL-21 relative to WT CD8⁺ T cells (Fig. 2G). The reduction in frequency of polarized CD8⁺ T cells correlated with reduced levels of mitochondrial Ca²⁺ in IL-2/IL-21–expanded T-cMCU KO CD8⁺ T cells compared to WT CD8⁺ T cells, as measured by Rhod2-AM staining and flow cytometry analysis (Supplementary Fig. S11E).

To test whether mitochondrial Ca²⁺ uptake through MCU contributed to the increased motility of IL-2/IL-21–expanded effector CD8⁺ T cells, WT and T-cMCU KO CD8⁺ T cells were activated, expanded with IL-2 alone or IL-2 plus IL-21, and then seeded for the transwell assay with medium alone. The enhanced motility of CD8⁺ T cells expanded with IL-2 plus IL-21 was abrogated for MCU-deficient CD8⁺ T cells (Fig. 2H). We also assessed the ability of IL-2/IL-21-expanded T-cMCU KO CD8⁺ T cells to produce IFNγ in response to re-stimulation with anti-CD3/CD28. There was no difference in IFNγ production between T-cMCU KO CD8⁺ T cells expanded in IL-2 plus IL-21 compared to WT CD8⁺ T cells expanded in the same conditions, as measured by intracellular staining and flow cytometry (Supplementary Fig. S11F). Therefore, the lack of MCU in effector CD8⁺ T cells selectively affected their motility triggered by IL-21 without interfering with proliferation or TCR-mediated activation.

Together, these results demonstrate that the presence of IL-21 during the expansion of effector CD8⁺ T cells with IL-2 increased their motility by reducing cytoskeleton stiffness through IL-21–mediated increase in mitochondrial Ca²⁺ levels. Moreover, while mitochondrial Ca²⁺ is essential for motility, it is dispensable for survival and TCR-mediated activation of effector CD8⁺ T cells.

The increase in mitochondrial Ca²⁺ levels triggered by IL-21 in effector CD8⁺ T cells is driven by its effect on MMP

To exclude that the effect of IL-21 on mitochondrial Ca²⁺ levels was due to an increase in mitochondrial content, we performed MitoTracker staining and flow cytometry analysis of effector CD8⁺ T cells after the second expansion with IL-2 or IL-2 plus IL-21. Similar MitoTracker staining was observed in both IL-2- and IL-2/IL-21–expanded CD8⁺ T cells (Fig. 3A). In addition, we also examined MitoTracker staining by confocal microscopy. No difference in mitochondrial content (based on MitoTracker area per cell) or mitochondria distribution within the cytosol was found between CD8⁺ T cells expanded with IL-2 plus IL-21 and those expanded with IL-2 alone (Supplementary Fig. S12A and S12B). To further demonstrate that mitochondrial content was not increased in IL-2/IL-21–expanded CD8⁺ T cells, we examined the levels of two mitochondrial proteins, electron transport chain Complex I subunit NDUFA9 and Complex III subunit UQCRC2, by Western blot analysis using whole cell extracts. There was no difference in the levels of either of the mitochondrial proteins between effector CD8⁺ T cells expanded with IL-2 and cells expanded with IL-2 plus IL-21 (Fig. 3B). Furthermore, no difference in cell size between IL-2–expanded and IL-2/IL-21–expanded CD8⁺ T cells was found, as determined by flow cytometry forward scatter analysis (Supplementary Fig. S13). Thus, the elevated mitochondrial Ca²⁺ levels in CD8⁺ T cells expanded in IL-2 plus IL-21 were not due to higher mitochondrial content or larger cell size.

Figure 3. Increased mitochondrial Ca²⁺ by IL-21 in effector CD8⁺ T cells is driven by higher MMP.

(A) WT CD8 cells were activated and expanded with IL-2 or IL-2 plus IL-21 for two expansions. Mitochondrial content was determined by MitoTracker Deep Red staining and flow cytometry. Representative histograms (left panel) and mean fluorescence intensity (MFI) (right panel, n=4 per group) are shown. (B) Western blot analysis for NDUFA9 and UQCRC2 using whole cell extracts from CD8 cells expanded with IL-2 or IL-2 plus IL-21. α-Tubulin levels are shown as a control. (C-E) MMP was determined by TMRE staining and flow cytometry in CD8 cells expanded with IL-2 or IL-2 plus IL-21. Representative histogram with High TMRE gate (C), percentage of cells with high TMRE fluorescent signal (as in C) (D), TMRE mean fluorescence intensity (MFI) (E) are shown, n=3. (F) CD8 cells were activated and expanded with IL-2 alone, IL-2/IL-21, IL-2/IL-7 or IL-2/IL-15 as in Figure 1A. After the 2nd expansion, MMP was determined by TMRE staining and flow cytometry. Percentage of cells with a high fluorescent signal (TMRE^high) is shown. (G) CD8 cells were activated and expanded in IL-2/IL-21 for two expansions, treated with CCCP (10 μM) or vehicle for 1.5 h, and mitochondrial Ca²⁺ was determined by Rhod-2AM staining and flow cytometry. (H) CD8 cells were activated, expanded in IL-2 plus IL-21, and were treated with CCCP (10 μM) or vehicle for the last 3 h of the 2nd expansion. Cells were cytospun, stained with rhodamine phalloidin and DAPI and examined by confocal microscopy. Quantification of cytoplasmic F-actin area/cell is shown, n=11 (Veh) and 13 (CCCP) number of cells. (I) CD8 cells were activated, expanded with IL-2 alone or IL-2 plus IL-21, and treated with vehicle or CCCP (10 μM) for the last 3 h of the 2nd expansion. Cells were then washed and placed into transwells for motility assay. Number of cells migrated into the wells is shown, n=3-4. For all, * denotes p < 0.05 and ns denotes not significant as determined by Unpaired t-test (D, E, H), One-Way ANOVA with multiple comparisons (F), or Two-way ANOVA with multiple comparisons (I). Mean ±SD is shown. All panels are representative of 2-5 independent experiments.

The uptake of Ca²⁺ into the mitochondrial matrix through MCU is driven by MMP (18). It is known that effector CD8⁺ T cells gradually lose their MMP over time during the expansion with IL-2 (33). Therefore, we investigated whether the higher mitochondrial Ca²⁺ levels present in IL-2/IL-21–expanded CD8⁺ T cells could be due to a higher MMP mediated by IL-21. Analysis of MMP by TMRE staining and flow cytometry showed higher MMP in CD8⁺ T cells expanded with IL-2 plus IL-21 compared to cells expanded with IL-2 alone, as determined by the percentage of cells with high TMRE (Fig. 3C and D), as well as by the mean fluorescence intensity of TMRE (Fig. 3E). In contrast to IL-21, the presence of IL-7 or IL-15 during the expansion of CD8⁺ T cells with IL-2 did not increase MMP (Fig. 3F). IL-21 also increased MMP in human CD8⁺ T cells expanded with IL-2 (Supplementary Fig. S14A and S14B). Thus, correlating with its effect on mitochondrial Ca²⁺ levels, IL-21 helped to sustain MMP during the expansion of effector CD8⁺ T cells with IL-2.

To further show the contribution of the high MMP to the increase in mitochondrial Ca²⁺ levels in effector CD8⁺ T cells expanded with IL-2 plus IL-21, we examined mitochondrial Ca²⁺ levels after treatment with CCCP, a mitochondrial uncoupler that dissipates the MMP (41). As determined by Rhod2-AM staining and flow cytometry analysis, mitochondrial Ca²⁺ levels in IL-2/IL-21–expanded CD8 cells were markedly reduced by CCCP treatment (Fig. 3G), showing the contribution of MMP to the level of mitochondrial Ca²⁺ in these cells. We then assessed the impact of CCCP treatment on cell polarity and cytoplasmic F-actin area of effector CD8⁺ T cells expanded with IL-2 plus IL-21. The treatment of IL-2/IL-21–expanded CD8⁺ T cells with CCCP for a short period of time caused areduction in the frequency of polarized cells (Supplementary Fig. 15A and 15B). Moreover, inhibition of MMP by CCCP treatment also increased the cytoplasmic F-actin fibers of IL-2/IL-21–expanded effector CD8⁺ T cells (Fig. 3H and Supplementary Fig. 15C). Thus, IL-21 reduced cytoskeleton stiffness of effector CD8⁺ T cells by sustaining MMP that, in turn, drove the increase in mitochondrial Ca²⁺ levels. To further show that MMP contributed to the superior motility of IL-2/IL-21–expanded effector CD8⁺ T cells, we performed the transwell assay. The enhanced motility of CD8⁺ T cells expanded with IL-2 plus IL-21 was abrogated by CCCP treatment (Fig. 3I).

Together, these data show that the presence of IL-21 during the expansion of effector CD8⁺ T cells with IL-2 sustained high MMP, leading to increased mitochondrial Ca²⁺ levels and superior motility.

Mitochondrial STAT3 mediates the effect of IL-21 on mitochondrial Ca²⁺ levels and motility of effector CD8⁺ T cells

While IL-2, IL-7, and IL-15 predominantly activate STAT5, IL-21 activates STAT3 (26, 42). The canonical function of STAT3 as a transcription factor is to mediate gene transcription following nuclear translocation (43). However, STAT3 has also been shown to translocate to the mitochondria, where it increases MMP by interacting with and enhancing the activity of Complex I of the electron transport chain (44-47). Phosphorylation of STAT3 at residue Ser⁷²⁷ is known to be required for mitochondrial STAT3 function, but it is not essential for STAT3 transcriptional activity (45, 48, 49). Since IL-21 increases MMP during the expansion of CD8⁺ T cells with IL-2, we examined the presence of STAT3 phosphorylated at S⁷²⁷ (phospho-STAT3-S⁷²⁷) by Western blot analysis of whole cell extracts. The levels of total STAT3 in effector CD8⁺ T cells expanded with IL-2 alone and those expanded with IL-2 plus IL-21 were comparable (Fig. 4A). However, high levels of phospho-STAT3-S⁷²⁷ were present only in CD8⁺ T cells expanded with IL-2 plus IL-21 (Fig. 4A). The selective presence of phospho-STAT3-S⁷²⁷ in IL-2/IL-21–expanded CD8⁺ T cells was further demonstrated by intracellular staining and flow cytometry analyses (Fig. 4B). While the presence of STAT3 in mitochondria in CD4⁺ T cells has been shown (21, 50), little is known about mitochondrial STAT3 in CD8⁺ T cells. We therefore examined the presence of STAT3 in mitochondria from effector CD8⁺ T cells expanded with IL-2 or with IL-2 plus IL-21 by Western blot analysis of mitochondrial extracts. High levels of total STAT3 were present only in the mitochondria of CD8⁺ T cells expanded with IL-2 plus IL-21 (Fig. 4C). In addition, mitochondrial STAT3 was phosphorylated at S⁷²⁷ in IL-2/IL-21–expanded CD8⁺ T cells (Fig. 4C). Thus, during the expansion of effector CD8⁺ T cells with IL-2, the presence of IL-21 promoted the accumulation and activation of mitochondrial STAT3.

Figure 4. Mitochondrial STAT3 is required for the enhanced MMP, mitochondrial Ca²⁺, and motility of IL-2/IL-21-expanded effector CD8⁺ T cells.

(A) WT CD8 cells were activated and expanded with IL-2 or IL-2 plus IL-21 for two expansions, and the levels of phospho-STAT3-S⁷²⁷ and total STAT3 were examined by Western blot analysis using whole cell extracts. β-actin was used as a control. (B) CD8 cells were expanded as in (A), and the levels of phospho-STAT3-S⁷²⁷ (blue and red histogram) were examined by intracellular staining and flow cytometry. Isotype control Ab in IL-2/IL-21-expanded cells is also shown (black histogram). (C) The levels of mitochondrial phospho-STAT3-S⁷²⁷ and total STAT3 in CD8 cells expanded as in (A) were examined by Western Blot analysis using mitochondrial extracts. GAPDH and mitochondrial protein UQCRC2 were used as controls. (D) CD8 cells from WT and STAT3-S⁷²⁷A KI mice were activated, expanded as in (A), and the levels of phospho-STAT3-S⁷²⁷ and total STAT3 were examined by Western blot analysis using whole cell extracts. α-tubulin was used as a control. (E-F) CD8 cells from WT and STAT3-S⁷²⁷A KI mice were activated, expanded as in (A) and examined for MMP by TMRE staining, n=3 (E), or for mitochondrial Ca²⁺ by Rhod2-AM staining and flow cytometry, n=3 (F). (G-H) WT CD8 cells were activated and expanded as in (A). During the last 6 hours of the 2nd expansion, cells were treated with STA-21 (10 μM) or vehicle and examined for mitochondrial Ca²⁺ by Rhod2-AM staining (G), or MMP by TMRE staining (H) and flow cytometry, n=3. (I-J) CD8 cells from WT and STAT3-S⁷²⁷A KI mice were activated and expanded with IL-2 plus IL-21 for two expansions. Frequency of polarized cells was assessed in live-cell brightfield microscope images of ICAM-Fc plates, n=3 images (I). Cells were cytospun, stained with rhodamine phalloidin for F-actin and analyzed by confocal microscopy. Quantification of cytoplasmic F-actin area was performed, n=11-13 cells/condition (J). (K) CD8 cells from WT and STAT3-S⁷²⁷A KI mice were activated and expanded with IL-2 or IL-2 plus IL-21 for two expansions. Number of migrated cells to the wells was determined by transwell assay, n=3. (L) WT CD8 cells were activated, expanded, and treated with STA-21 (10 μM) or vehicle as in G, washed and transwell motility assay was performed, n=4. * denotes p <0.05 and ns denotes not significant as determined by Two-Way ANOVA with multiple comparisons (E, F, K) or Unpaired t-test (G-J, L). Mean ±SD is shown. All panels are representative of 2-3 independent experiments.

To assess whether the high MMP in CD8⁺ T cells expanded with IL-2 plus IL-21 was mediated by mitochondrial STAT3, we used CD8⁺ T cells from STAT3-S⁷²⁷A KI (STAT3 KI) mice where STAT3 Ser⁷²⁷ is replaced by Ala to prevent its phosphorylation (29). CD8⁺ T cells from STAT3 KI and WT mice were isolated, activated, and expanded in IL-2 or IL-2 plus IL-21. No difference in proliferation or survival was detected between WT and STAT3 KI CD8⁺ T cells after two expansions with either IL-2 alone or IL-2 plus IL-21 (Supplementary Fig. S16A). There was also no difference in the phenotype based on the expression of CD44, CCR7 and CD127 (Supplementary Fig. S16B). Loss of phosphorylation of STAT3 at Ser⁷²⁷ in STAT3 KI CD8⁺ T cells was confirmed by Western blot analysis using whole cell extracts from effector CD8⁺ T cells (Fig. 4D). The levels of total STAT3 were not affected in STAT3 KI CD8⁺ T cells (Fig. 4D). Moreover, comparable levels of STAT3 phosphorylated at Tyr⁷⁰⁵ were present in WT and STAT3 KI CD8⁺ T cells expanded with IL-2 and IL-21 (Supplementary Fig. S16C). We then examined MMP by TMRE staining in WT and STAT3 KI CD8⁺ T cells expanded with IL-2 or IL-2 plus IL-21. MMP was markedly reduced in IL-2/IL-21–expanded STAT3 KI CD8⁺ T cells relative to the MMP in IL-2/IL-21–expanded WT CD8⁺ T cells, based on percentage of TMRE^high cells (Fig. 4E and Supplementary Fig. S16D), as well as TMRE mean fluorescence intensity (Supplementary Fig. S16E). Thus, IL-21 helped to maintain high MMP levels during effector CD8⁺ T-cell expansion through mitochondrial STAT3.

Since we showed that increased mitochondrial Ca²⁺ levels of IL-2/IL-21–expanded CD8 cells are driven by their higher MMP, we investigated whether the effect of IL-21 on mitochondrial STAT3 could be the mechanism by which this cytokine regulates mitochondrial Ca²⁺. We measured mitochondrial Ca²⁺ in WT and STAT3 KI CD8⁺ T cells expanded with IL-2 or IL-2 plus IL-21. STAT3 KI CD8⁺ T cells expanded with IL-2 plus IL-21 exhibited a marked reduction in mitochondrial Ca²⁺ levels compared to IL-2/IL-21–expanded WT CD8⁺ T cells (Fig. 4F and Supplementary Fig. S16F), indicating that the regulation of mitochondrial Ca²⁺ by IL-21 during the expansion of effector CD8⁺ T cells required functional mitochondrial STAT3. To further demonstrate that the regulation of mitochondrial Ca²⁺ by IL-21 was independent of the canonical function of STAT3 as a transcription factor, we investigated the effect of STA-21, a well-established pharmacological inhibitor of STAT3 transcriptional activity (51). WT CD8⁺ T cells were activated, expanded with IL-2 plus IL-21, and treated with STA-21 or vehicle for the last 6 h of the second expansion. Analysis of mitochondrial Ca²⁺ by flow cytometry showed no effect of STA-21 (Fig. 4G and Supplementary Fig. S17A). Similarly, no differences in MMP were observed between IL-2/IL-21–expanded CD8⁺ T cells treated with STA-21 and those treated with vehicle (Fig. 4H and Supplementary Fig. S17B). Thus, the IL-21–driven increase in mitochondrial Ca²⁺ in effector CD8⁺ T cells during their expansion was mediated by mitochondrial STAT3 independently of STAT3 transcriptional activity.

We therefore investigated whether mitochondrial STAT3 was also needed for the reduced cytoskeleton stiffness and enhanced motility of effector CD8⁺ T cells driven by IL-21. CD8⁺ T cells from WT and STAT3 KI mice were isolated, activated, and expanded with IL-2 or IL-2 plus IL-21. Morphological examination revealed a reduced percentage of polarized STAT3 KI CD8⁺ T cells after the second expansion with IL-2 plus IL-21 compared to the high frequency found for WT CD8⁺ T cells (Fig. 4I). Accordingly, analysis of cytoplasmic F-actin also showed higher content of cytoplasmic fibers in STAT3 KI CD8⁺ T cells expanded in IL-2 plus IL-21 compared to WT CD8⁺ T cells (Fig. 4J and Supplementary Fig. S18). We then examined the motility of WT and STAT3 KI CD8⁺ T cells following the second expansion with IL-2 alone or IL-2 plus IL-21 by transwell assay. In contrast to its effect on WT CD8⁺ T cells, IL-21 failed to increase the motility of STAT3 KI CD8⁺ T cells (Fig. 4K). We also assessed the contribution of STAT3 transcriptional activity to the motility of effector CD8⁺ T cells expanded in the presence of IL-21 by treating WT CD8⁺ T cells with STA-21 or vehicle during the last 6 h of the second expansion. No difference in motility was observed between WT CD8⁺ T cells treated with STA-21 or vehicle (Fig. 4L). Thus, the enhanced motility of effector CD8⁺ T cells expanded with IL-2 in the presence of IL-21 relative to the motility of IL-2–expanded CD8⁺ T cells required mitochondrial STAT3 function but not transcriptional activity.

To further demonstrate the contribution of mitochondrial STAT3 to in vivo motility of effector CD8⁺ T cells expanded with IL-2 plus IL-21, CD8⁺ T cells were isolated from WT and STAT3 KI mice, activated, expanded with IL-2 plus IL-21 for two expansions, fluorescently labeled with CFSE or CTV, respectively, and i.v. co-administered together at 1:1 ratio into WT mice. After 24 h, LNs were harvested, and the motility of the transferred cells within intact LNs was measured using time-lapse imaging with two-photon microscopy. STAT3 KI effector CD8⁺ T cells on average traveled shorter paths (Supplementary Fig. S19A), with lower mean square displacement (Supplementary Fig. S19B), lower total displacement (Supplementary Fig. S19C), and lower average speed (Supplementary Fig. S19D and Supplementary Videos S7 and S8), compared to WT effector CD8⁺ T cells. Therefore, mitochondrial STAT3 was critical for the enhanced in vivo motility of effector CD8⁺ T cells expanded with IL-2 and IL-21.

Together, our data show that IL-21 increased the motility of effector CD8⁺ T cells by sustaining a high MMP through mitochondrial STAT3, which in turn enhanced mitochondrial Ca²⁺ levels and reduced cytoskeleton stiffness.

High motility of effector CD8⁺ T cells expanded in the presence of IL-21 enhances their antitumor efficacy in vitro

Cell motility is a key factor for effector CD8⁺ T cells to search for and reach their target cancer cells expressing specific antigens within solid tumors. While IL-2 is the most commonly used cytokine for expanding adoptive T-cell therapies (52), our data above show that effector CD8⁺ T cells expanded with IL-2 alone exhibit limited motility. We therefore assessed whether the high motility of effector CD8⁺ T cells expanded with IL-2 plus IL-21 resulted in a greater efficacy of killing antigen-specific cancer cells in vitro. We used CD8⁺ T cells from OT-I transgenic mice, which express a specific TCR that recognizes chicken OVA peptide presented by MHC class I (53), and B16 melanoma cells expressing OVA (B16-OVA cells) as a foreign antigen (54). OT-I CD8⁺ T cells were activated and expanded with IL-2 or IL-2 plus IL-21. After the second expansion, OT-I cells were co-cultured with B16-OVA melanoma cells. Effector OT-I CD8⁺ T cells expanded with IL-2 plus IL-21 exhibited superior killing activity of B16-OVA melanoma cells relative to OT-I CD8⁺ T cells expanded with IL-2 alone (Fig. 5A). We also performed similar experiments, but instead of directly adding OT-I CD8⁺ T cells to the B16-OVA melanoma cell cultures, IL-2- or IL-2/IL-21-expanded OT-I CD8⁺ T cells were placed in transwells above melanoma cell cultures. Thus, only OT-I CD8⁺ T cells that migrated through the transwells could kill the tumor cells. Melanoma killing by IL-2/IL-21–expanded OT-I CD8⁺ T cells was increased relative to killing by OT-I CD8⁺ T cells expanded only with IL-2 (Fig. 5B), indicating that the enhanced motility and ability to move through the transwell of effector CD8⁺ T cells expanded with IL-2 plus IL-21 translated into superior tumor cell killing efficacy.

Figure 5. Enhanced motility of effector CD8⁺ T cells expanded with IL-2 and IL-21 increases their melanoma cell killing efficacy in vitro.

(A) CD8 cells from OT-I mice were activated, expanded in IL-2 or IL-2 plus IL-21 for two expansions, washed and added to B16-OVA cell cultures at E:T (Effector: Target) =1. 24 h later, live B16-OVA cells were determined by trypan blue staining, n=3-4. (B) OT-I CD8 cells were activated, expanded and washed as in (A), and placed into transwell chambers over B16-OVA cell cultures at E:T=0.5. 24 h later, live B16-OVA cells were determined, n=3. (C) OT-I CD8 cells were activated, expanded and washed as in (A), and added to B16-OVA cell cultures that were at a low (40%) and high (100%) confluency of B16-OVA cells. 24 h later, live B16-OVA cells were determined, n=3-4. (D) OT-I CD8 cells were activated, expanded and washed as in (A), and added to cultures with a mix (1:1) of B16-OVA cells and B16 cells. 24 h later, live B16 and B16-OVA cells were determined, n=4. (E) IL-2- and IL-2/IL-21-expanded OT-I CD8 cells were pretreated with vehicle or CGP (10 μM) for 7 h, washed, and added to B16-OVA cell cultures. 24 h later, live B16-OVA cells were determined, n=4. (F-G) OT-I CD8 cells were activated, expanded and washed as in (A), and restimulated with anti-CD3/CD28 Abs for 20 h. IL-2 production, n=3 (F) and TNFα production, n=2 (G) were evaluated by intracellular staining and flow cytometry. (H) CD8 cells were isolated from OT-I and T-cMCU KO OT-I mice, activated, expanded and washed as in (A), and added to B16-OVA cell cultures. 24 h later, live B16-OVA cells were counted, n=4. * denotes p < 0.05 and ns denotes not significant as determined by Unpaired t-test (A-B, D) or Two-Way ANOVA with multiple comparisons (C, E-H). Mean ±SD is shown. All panels are representative of 2-5 independent experiments.

To show that the superior killing of cancer cells by OT-I CD8⁺ T cells expanded with IL-2 plus IL-21 was mediated by enhanced motility and a greater capacity to encounter target cells, a different confluency of B16-OVA cells was tested in the co-cultures. With low confluency of the B16-OVA melanoma cells in the cultures, OT-I CD8⁺ T cells need to travel longer distances to reach the sparse cancer cells. Different numbers of B16-OVA cells were plated, and when they reached approximately 40% or 100% confluency, OT-I CD8⁺ T cells expanded with IL-2 or IL-2 plus IL-21 were added to the co-cultures. IL-2–expanded OT-I cells were as competitive as the IL-2/IL-21–expanded cells in killing B16-OVA cells at high cancer cell confluency (Fig. 5C). However, at low B16-OVA cell confluency, IL-2/IL-21–expanded OT-I cells retained most of their killing efficacy, but the killing activity of IL-2–expanded OT-I cells was abrogated (Fig. 5C).

To further demonstrate the contribution of the enhanced motility of IL-2/IL-21–expanded OT-I CD8⁺ T cells to their superior killing of tumor cells, we employed an alternative in vitro experimental design that more closely represented the physiological tumor context. We plated B16 melanoma cells lacking expression of OVA together with the same number of B16-OVA melanoma cells to obtain high confluency melanoma cell cultures where the OVA-expressing cancer cells were sparse. IL-2– or IL-2/IL-21–expanded OT-I CD8⁺ T cells were added to the B16/B16-OVA cell mixed cultures. OT-I cells expanded with IL-2 plus IL-21 exhibited a superior killing efficacy compared to those expanded with IL-2 alone (Fig. 5D). Together, these results demonstrated the competitive advantage of IL-2/IL-21–expanded antigen-specific CD8 cells in killing tumor cells, particularly when they had to travel to search for and encounter antigen-specific tumor cells.

We showed above (Supplementary Figure S5) that, despite their superior motility, CD8⁺ T cells expanded with IL-2 plus IL-21 responded similarly to anti-CD3/CD28 stimulation (where cell motility is dispensable) compared to CD8⁺ T cells expanded with IL-2 only. We verified that the response of IL-2/IL-21–expanded OT-I CD8⁺ T cells to anti-CD3/anti-CD28 stimulation was comparable to the response of IL-2–expanded OT-I CD8⁺ T cells, as determined by the production of IL-2 and IFNγ (Supplementary Fig. S20A and S20B). Thus, the superior antitumor killing activity of IL-2/IL-21–expanded OT-I cells was not a result of increased TCR stimulation.

Considering that we showed that mitochondrial Ca²⁺ was dispensable for proliferation and TCR-mediated activation but essential for the motility of CD8⁺ T cells (Fig. 2F and H, Supplementary Fig. S10 and Supplementary Fig. S11), we interfered with mitochondrial Ca²⁺ levels to assess their contribution to the enhanced killing activity observed for IL-2/IL-21–expanded OT-I cells. OT-I CD8⁺ T cells were activated, expanded with IL-2 or IL-2 plus IL-21, and treated with CGP or vehicle for 7 h at the end of the second expansion. After extensive washes, OT-I cells were added to B16-OVA cultures. Inhibition of mitochondrial Ca²⁺ efflux with CGP prevented the superior killing efficacy of IL-2/IL-21–expanded OT-I CD8⁺ T cells (Fig. 5E). We verified that treatment with CGP did not interfere with TCR signaling and activation of the IL-2/IL-21–expanded OT-I cells. After the second expansion with IL-2 alone or IL-2 plus IL-21, OT-I CD8⁺ T cells were restimulated with anti-CD3/anti-CD28 in the presence of vehicle or CGP and cytokine production was assessed by staining and flow cytometry. No difference in production of IL-2 (Fig. 5F) or TNFα (Fig. 5G) was observed.

We also examined the ability of MCU-deficient effector CD8⁺ T cells to kill tumor cells. We crossed OT-I mice with T-cMCU KO mice to generate T-cMCU KO OT-I mice. Lack of MCU did not affect the development of OT-I CD8⁺ T cells (Supplementary Fig. S21A and S21B). WT and T-cMCU KO OT-I CD8⁺ T cells were isolated, activated, and expanded with IL-2 or IL-2 plus IL-21. We then examined the tumor-killing activity of these cells in co-cultures with B16-OVA melanoma cells. The presence of IL-21 during the expansion of MCU-deficient OT-I CD8⁺ T cells failed to enhance their killing activity against melanoma cells (Fig. 5H).

In addition, we also tested the ability of effector STAT3 KI CD8⁺ T cells to kill antigen-specific tumor cells. Total spleen cells from WT and STAT3 KI mice were activated with the OVA SIINFEKL peptide (N4 peptide) known to be presented by MHC-I and recognized with high affinity by the TCR present in OT-I CD8⁺ T cells (55) and IL-2 to facilitate the initial expansion. After 5 days, those activated CD8⁺ T cells were further expanded with IL-2 alone or IL-2 plus IL-21, and their tumor-killing activity was assayed in co-cultures with B16-OVA melanoma cells. Similar to WT OT-I CD8⁺ T cells, polyclonal WT CD8⁺ T cells activated with the N4 peptide and expanded with IL-2 plus IL-21 also displayed superior killing of B16-OVA cells relative to the killing by WT CD8⁺ T cells expanded with IL-2 alone (Supplementary Fig. S22). However, the antigen-specific killing activity of STAT3 KI CD8⁺ T cells expanded with IL-2 plus IL-21 was inferior compared to the killing activity of IL-2/IL-21–expanded WT CD8⁺ T cells (Supplementary Fig. S22).

Together, the data presented here show that the superior in vitro antitumor killing activity of effector CD8⁺ T cells expanded with IL-2 plus IL-21 was due to their enhanced motility.

The enhanced motility of TCR-specific CD8⁺ T cells expanded in the presence of IL-21 results in superior efficacy against melanoma in vivo

TCR-T cell therapy, in which adoptively transferred CD8⁺ T cells express a specific TCR that recognizes a tumor-cell antigen in the context of MHC class I is an emerging therapeutic modality for treatment of solid tumors (56). However, one weakness of this therapy is the limited motility of the in vitro–expanded effector CD8⁺ T cells transferred, which limits their ability to move throughout the tumor microenvironment and find antigen-specific cancer cells. Since IL-2 is the primary growth factor used to expand TCR-T cells for cancer treatment, we investigated whether the addition of IL-21 to IL-2 during expansion could improve the efficacy of TCR-T cell therapy for melanoma in vivo. Similar to the in vitro experimental design, a mixture of B16 cells expressing OVA and lacking OVA was implanted on the flank of WT mice. OT-I CD8⁺ T cells were activated, expanded with IL-2 alone or IL-2 plus IL-21, and adoptively transferred into B16/B16-OVA melanoma-bearing mice. Tumor size and survival of the mice were followed over time. Survival of mice treated with OT-I cells expanded with IL-2 plus IL-21 was superior to the survival of the cohort of mice treated with IL-2–expanded OT-I cells (Fig. 6A). Increased survival was associated with a marked delay in tumor growth in the mice that had received OT-I cells expanded with IL-2 plus IL-21 (Fig. 6B).

Figure 6. Enhanced motility of effector CD8⁺ T cells expanded in the presence of IL-21 confers superior antitumor efficacy in vivo.

(A-B) CD8 cells from OT-I mice were activated and expanded in IL-2 or IL-2 plus IL-21. After two expansions, IL-2-expanded or IL-2/IL-21-expanded OT-I cells were separately transferred into B16/B16-OVA tumor-bearing mice. Survival (A) and tumor size (B) were followed. n=6 (IL-2) and 7 (IL-2/IL-21) number of mice. (C) CD8 OT-I cells activated and expanded as in (A) were labeled with CFSE (IL-2) or Cell Trace Violet (IL-2/IL-21) and i.v. co-injected at equal ratio into B16-OVA tumor-bearing mice. After 2 days, tumors were harvested, digested, and the frequency of transferred OT-I cells was assessed by flow cytometry, n=4. (D-F) CD8 OT-I cells activated and expanded as in (A) were labeled with CFSE (IL-2) or Cell Trace Violet (IL-2/IL-21) and i.v. co-injected at equal ratio into B78mChOVA tumor-bearing mice. After 2 days, tumors were harvested and cell motility was measured by two-photon microscopy over 30 mins. Representative image showing trajectories for IL-2-expanded cells (green) and IL-2/IL-21-expanded cells (magenta) (D), mean square displacement per cell (E), total displacement per cell (F), and average speed per cell (G) measured with Imaris software. n=17 (IL-2) and n=8 (IL-2/IL-21) number of cells. (H-I) WT OT-I and T-cMCU KO OT-I CD8 cells were activated, expanded in IL-2/IL-21, and adoptively transferred into B16/B16-OVA-tumor-bearing mice as in (A). Survival (H) and tumor size (I) were followed. n=6 mice per group. * denotes p < 0.05 and ns denotes not significant as determined by Unpaired t-test (C, F-G), Log-rank Mantel-Cox test (A, H), Mixed-effects analysis (B, I), or Two-way ANOVA (E). Mean ±SEM is shown (B, E, I). Maximum, minimum, median, 25^th percentile and 75^th percentile are shown (F, G). All panels are representative of 2 independent experiments.

To assess whether the superior efficacy of IL-2/IL-21–expanded OT-I CD8⁺ T cells could be due to an increased recruitment to the tumor, OT-I CD8⁺ T cells were isolated, activated, expanded in IL-2 or IL-2 plus IL-21, labeled with CFSE or CTV, and adoptively co-transferred (1:1 ratio of IL-2–expanded versus IL-2 plus IL-21–expanded cells) into B16-OVA tumor-bearing mice. Melanoma tumors were harvested two days later, and the relative frequency of transferred OT-I cells within the tumors was measured by flow cytometry. No difference in relative frequency of OT-I cells expanded with IL-2 and those expanded with IL-2 plus IL-21 was found (Fig. 6C), indicating that trafficking to and penetrance of the tumor was not improved when TCR-specific CD8⁺ T cells were expanded with IL-21 in combination with IL-2. We then examined the motility of the adoptively transferred OT-I CD8⁺ T cells within the tumor. We used B78mChOVA melanoma cells derived from B16 cells that were engineered to lack melanin and express both mCherry and OVA antigen, making the tumors more suitable for imaging (57). OT-I CD8⁺ T cells were isolated, activated, expanded with IL-2 or IL-2 plus IL-21, fluorescently labeled, and co-transferred into B78mChOVA-tumor-bearing mice. Two days later, tumors were harvested, and intratumoral motility was examined by two-photon microscopy. OT-I cells expanded with IL-2 plus IL-21 traveled longer distances compared to IL-2–expanded OT-I cells, as determined by increased mean square displacement as well as total displacement (Fig. 6D-F and Supplementary Video S9). In addition, OT-I cells expanded with IL-2 plus IL-21 moved faster within the tumor as determined by the average speed (Fig. 6G and Supplementary Video S9). Thus, expansion of TCR-specific CD8⁺ T cells in the presence of IL-21 in addition to IL-2, resulted in greater intratumoral motility of the transferred cells, which could be reflected in superior antitumor efficacy of the treatment.

To further demonstrate that the superior antitumor efficacy of OT-I CD8⁺ T cells expanded with IL-2 plus IL-21 was due to their competitive motility, we used OT-I CD8⁺ T cells from OT-I T-cMCU KO mice, since we showed above that mitochondrial Ca²⁺ was essential for the motility but not for the proliferation or activation of IL-2/IL-21–expanded CD8⁺ T cells. OT-I and T-cMCU KO OT-I CD8⁺ T cells were isolated, activated, expanded with IL-2 plus IL-21 and independently administered to B16/B16-OVA tumor-bearing mice. Survival of B16/B16-OVA tumor-bearing mice that were treated with IL-2/IL-21–expanded T-cMCU KO OT-I cells was markedly reduced relative to the group of B16/B16-OVA tumor-bearing mice that received IL-2/IL-21–expanded WT OT-I cells as treatment (Fig. 6H). Consistent with their poor survival, a substantial increase in tumor growth rate was observed in the cohort of mice receiving IL-2/IL-21–expanded T-cMCU KO OT-I cells (Fig. 6I). To rule out any differences in trafficking to or penetrance of the tumor between WT and T-cMCU KO OT-I cells, CD8⁺ T cells were isolated from WT OT-I and T-cMCU KO OT-I mice, activated, expanded in IL-2 plus IL-21, labeled with CFSE or CTV, and adoptively co-transferred at a 1:1 ratio into B16-OVA tumor-bearing mice. Tumors were harvested two days later, and the presence of WT and T-cMCU KO OT-I cells within the tumors was measured by flow cytometry. No difference in frequency of WT and T-cMCU KO OT-I cells was found (Supplementary Fig. S23) indicating that the reduced antitumor response of mice receiving T-cMCU KO OT-I cells was due to reduced cell motility and not reduced cell recruitment to the tumor.

Together, these studies show that the antitumor efficacy of TCR-T cell therapy for solid tumors (e.g. melanoma) can be improved if IL-21 is present during the expansion of TCR-specific CD8⁺ T cells with IL-2 prior to the administration, by enhancing CD8⁺ T-cell intratumor motility and ability to find antigen-specific cancer cells.

DISCUSSION

Effector CD8⁺ T cells are a primary component of the various adoptive T-cell therapies for treating hematologic and solid malignancies. These therapies involve isolating T cells from peripheral blood of the patients, genetically modifying them to express a receptor, and extensively expanding them ex vivo, most commonly with IL-2, before administering them to the patient. Trafficking of the transferred T cells to tumors is highly relevant and has been extensively investigated (9). However, another essential aspect of CD8⁺ T-cell antitumor function that is not well appreciated and remains under-investigated even in the context of T-cell therapy is cell motility, defined as random walk and exploratory spread movement, which is required for effector CD8⁺ T cells to search for their specific cell targets within solid tumors. If only a small percentage of effector CD8⁺ T cells that are transfused to the patient are motile and able to search for target cells within the tumor, the efficacy will be poor regardless of their cytotoxicity. Effector CD8⁺ T-cell motility is also relevant for adoptive T-cell therapies because of the requirement for a large in vitro expansion of these effector cells with IL-2 to obtain sufficient numbers for transfusion, and the resulting loss of cell polarity due to their expansion. Herein, we have shown that the addition of IL-21 during the expansion of effector CD8⁺ T cells with IL-2 selectively enhances the motility of these cells without interfering with their proliferation or activation, and this enhanced motility is translated into a superior killing efficacy against melanoma cells in vitro and in vivo.

Unlike non-hematopoietic cells that form the bulk of most tissues, naive and activated T cells largely display a rounded morphology, which is supported by stiff F-actin fibers across the cytoplasm. Motile T cells need to reduce their cytoskeleton stiffness to acquire a polarized morphology with the formation of lamellipodia composed of branched actin at their leading edge and a uropod protrusion in their rear (58). Mitochondrial Ca²⁺ is a critical regulator of the actin cytoskeleton since it augments cytoskeleton dynamics and reduces cytoskeleton stiffness (14, 16, 21, 38). Alterations in mitochondrial Ca²⁺ levels have been shown to impair migration of different cell types including endothelial cells and embryonic stem cells (16, 59), as well as sperm motility (60). In our study, we have shown that effector CD8⁺ T cells expanded with IL-2 have higher density of cytoplasmic F-actin fibers maintaining their rounded morphology than effector CD8⁺ T cells expanded with IL-2 in the presence of IL-21. Reduced cytoskeleton stiffness in effector CD8⁺ T cells expanded with IL-2 plus IL-21 allows these cells to acquire a polarized morphology. Moreover, we found that IL-21 increases mitochondrial Ca²⁺ levels in effector CD8⁺ T cells during expansion, and that mitochondrial Ca²⁺ is required for IL-21 to reduce cytoskeleton stiffness and enhance CD8⁺ T-cell motility. Previous studies have shown that mitochondrial Ca²⁺ is dispensable for CD8⁺ T-cell proliferation and cytotoxicity (20). Our results here confirm that mitochondrial Ca²⁺ is dispensable for the activation, proliferation, and cytokine production of effector CD8⁺ T cells expanded with IL-2 plus IL-21, when motility is not required. Thus, the selective ability of IL-21 to increase mitochondrial Ca²⁺ levels confers this cytokine with the function of increasing the motility of effector CD8⁺ T cells, complementing the proliferative capacity provided by IL-2.

IL-21 signals through the JAK/STAT pathway (61). IL-21 primarily activates the transcription factor STAT3, which induces the expression of various genes, such as Prdm1, Bcl3, c-myc, and Junb (62-64). Our question was how IL-21 plays a role in regulating mitochondrial Ca²⁺ levels in effector CD8⁺ T cells. Beyond its classical role in gene transcription, STAT3 also localizes to mitochondria (despite lacking a mitochondrial targeting sequence) where it increases MMP by acting as a positive regulator of Complex I of the ETC through interaction with Grim19 (44, 45, 65-67). While the function of STAT3 as a transcription factor is well established in CD8⁺ T cells (36, 68, 69), to our knowledge, no previous studies have reported the presence of STAT3 in mitochondria of CD8⁺ T cells or its potential role within this compartment in CD8 cells. Our studies here show the selective presence of STAT3 in mitochondria of effector CD8⁺ T cells when these cells are expanded with IL-2 and IL-21. Mitochondrial STAT3 is dispensable for the proliferation and survival of effector CD8⁺ T cells, as expected since IL-2 does not activate STAT3. However, we reveal here that mitochondrial STAT3 mediates the increase in mitochondrial Ca²⁺ levels triggered by IL-21 in effector CD8⁺ T cells and, as a result, mitochondrial STAT3 is also critical for IL-21–induced motility of effector CD8⁺ T cells. In addition, unlike most other functions of STAT3 (36, 68, 70), the transcriptional activity of STAT3 is not essential for IL-21 to promote increased levels of mitochondrial Ca²⁺ and subsequent motility of effector CD8⁺ T cells.

Adoptive T-cell therapies (e.g. CAR-T cell therapy) have proven relatively successful for hematological malignancies, where overall response rates are high (71, 72), but their use for the treatment of solid tumors has been more challenging. TCR-T cell therapy is an emerging adoptive T-cell therapy in which T cells are genetically engineered to express a specific TCR that recognizes cancer cell–derived peptides presented on MHC class I, known to be selectively recognized by CD8⁺ T cells (56). The first TCR-T cell therapy was FDA approved for the treatment of synovial sarcoma (5). However, only a fraction of cancer cells in solid tumors can express the target antigen, process it into specific peptides, and present these peptides in the context of MHC class I to the engineered TCR-expressing CD8⁺ T cells. Thus, the motility of the transferred TCR-expressing CD8⁺ T cells is critical for these cells to find the MHC-I/peptide–expressing cancer cells within the tumor and ensure the success of the TCR-T cell therapy. Our in vitro and in vivo models, which utilize a mixture of antigen-expressing and antigen-nonexpressing melanoma cells, mimic the need for CD8⁺ T cells to travel to their antigen-expressing target cells found in solid tumors. We show here that the enhanced motility of TCR-expressing effector CD8⁺ T cells expanded with IL-2 in the presence of IL-21 results in an in vitro superior killing of antigen-specific melanoma cells when the cancer cells are sparse relative to the poorly motile CD8⁺ T cells expanded in IL-2 alone. Moreover, we show that the enhanced motility of TCR-expressing CD8⁺ T cells generated in the presence of IL-21 results in superior efficacy of this TCR-T cell therapy against melanoma in vivo. Previous studies have investigated the effect of overexpressing IL-21 or IL-21R in T cells used for T-cell therapy in the context of survival or activation (73, 74). We reveal here that the addition of IL-21 during the manufacturing of TCR-T cell therapy is beneficial for the success of this therapy in solid tumors, not simply by increasing survival or activation of the transferred CD8⁺ T cells (25, 75) but by enhancing their motility and the chances of finding their antigen-specific targets.

Overall, our studies show that IL-21 confers enhanced motility on effector CD8⁺ T cells through an unconventional mechanism involving mitochondrial STAT3 and mitochondrial Ca²⁺. Moreover, our studies reveal that expanding effector CD8⁺ T cells with a combination of IL-2 and IL-21 is a viable strategy to enhance their intratumoral motility and improve the efficacy of adoptive TCR-T cell therapies for solid cancers. The translational impact of these studies is clear, and it can be applied to other adoptive T-cell therapies.

SYNOPSIS.

Killing of cancer cells in solid tumors requires T cells randomly walk and search for their target. The authors reveal that increasing mitochondrial Ca²⁺ levels enhances effector CD8⁺ T-cell motility, improving adoptive T cell–therapy efficacy against solid tumors.