Intracellular K⁺ Limits T-cell Exhaustion and Preserves Antitumor Function

The past 20 years of translational oncology provide a framework in which the ability to eradicate cancer has increased in direct relation to our understanding of T-cell biology (1). Although T cells infiltrate cancers, only a small proportion of patients that receive immunotherapy achieve durable regression of disease, owing to tumor-induced suppression of T-cell function (2–4). Following T-cell receptor (TCR) engagement, naïve T cells (T_N) progressively differentiate into stem cell memory (T_SCM), central memory (T_CM), effector memory (T_EM), and terminally differentiated exhausted T cells (T_Ex). As T cells differentiate along this continuum, transcriptional, epigenetic, and metabolic changes drive the acquisition of effector functions (IFNγ production, cytolysis) in T_EM cells, along with the coincident loss of stem-like behaviors (multipotency, persistence, and self-renewal) seen in T_SCM and T_CM cells (5). Within cancers, repetitive TCR stimulation drives T_EM cells to differentiate into T_Ex cells, with compromised effector functions and loss of stem-cell like behavior. In addition to T_Ex cell hyporesponsiveness, direct suppression of TCR signaling (i.e., PD-L1, CTLA-4) can drive T-cell intratumoral dysfunction (1, 2). We previously reported that elevated extracellular potassium (↑[K⁺]_e) within tumors directly suppresses TCR-induced signal transduction and effector function (6–8). In the current work, we aimed to imbue tumor-specific T cells with resistance to ↑[K⁺]_e by reprogramming K⁺ transport to result in a lower intracellular K⁺ ([K⁺]_i).

To understand the underpinnings of T-cell K⁺ physiology, we performed whole-transcriptomic analysis of T cells in varied states of activation and differentiation. Although our efforts and those of others have focused on the role of the K⁺ channels K_v1.3 and K_Ca3.1, we also found that transcripts encoding a number of other K⁺ channels and pumps were differentially abundant across T-cell populations (9). We elected to focus on the Na⁺/K⁺ ATPase, owing to its established contribution to [K⁺]i, resting membrane potential (i.e., V_m), and robust expression (10). The Na⁺/K⁺ ATPase exports three sodium ions and imports two potassium ions, producing a cytoplasm that is hyperkalemic and hyponatremic with a V_m of approximately −60 mV compared with the extracellular space (11). Despite depolarizing the resting V_m and compromising Ca²⁺ influx, genetic disruption of the Na⁺/K⁺ ATPase produced spontaneous activation within TCR and Akt–mTOR signaling cascades. Ongoing signal transduction resulted in T_Ex cell formation, mirroring other instances of tonic T-cell signaling. Mechanistically, we found Atp1a1-deficient T cells to be in a state of stress response, with elevated levels of reactive oxygen species (ROS). Treatment with the antioxidant N-acetylcysteine (NAC) prevented ROS accumulation and T_Ex cell formation in vitro. Yet, the driving source of dysfunction in Na⁺/K⁺ ATPase deficient T cells was a depletion of [K⁺]_i. Provision of [K⁺]_e raised [K⁺]_i levels and prevented ROS accumulation and phenotypic exhaustion. Consistent with their T_Ex cell phenotype, tumor-specific T cells lacking Atp1a1 had compromised in vivo persistence and antitumor activity. These findings indicate that a modicum of [K⁺]_i is required in T cells to prevent unregulated signal transduction in TCR and Akt–mTOR signaling, resulting in T-cell exhaustion. Future efforts will focus on understanding how [K⁺]_i regulates ROS and providing T cells the abilities to weather the stressors of the tumor microenvironment, including pathologic ion concentrations. A deeper understanding of T-cell ion transport will advance cancer immunotherapy by informing the development and application of pharmacologic and genetic interventions that prevent T-cell exhaustion, while also augmenting intratumoral effector functions.

All mouse experiments were approved by the Animal Care and Use Committee at OHSU (Department of Comparative Medicine). All infectious agents were approved by the OHSU Institutional Biosafety Committee. Deidentified healthy donors (HD) were recruited and enrolled by the Oregon Clinical and Translational Research Institute under the Research Volunteer Repository program.

All mice were obtained from The Jackson Laboratory. C57BL/6 mice (Cat. 000664) of 6–8 weeks of age were used as recipient hosts for adoptive transfer, unless otherwise indicated. OT-1 Ly5.1 transgenic mice were used for adoptive cell transfer (ACT) and viral kinetics experiments. To obtain OT-1 Ly5.1 mice (CD45.1/1⁺), we crossed OT-1 Ly5.2 (CD45.2/2⁺) mice (C57BL/6-Tg (TcraTcrb)1100Mjb/J; Cat. 003831) with Ly5.1 (CD45.1/1+) mice (B6.SJL-PtprcaPepcb/BoyJ; 002014). This resulted in OT-1 mice heterogenous for Ly5.1 (CD45.1/2⁺), which were further crossed to obtain OT-1 Ly5.1 homozygotes. We bred OT-1 mice with Rag2^−/− mice (B6.Cg-Rag2tm1.1gGn/J; Cat. 008449) to obtain OT-1 Rag2^−/− progeny. All mice were maintained under specific pathogen-free conditions.

B16-OVA was generated from parental B16-F10 (obtained from the ATCC, Cat no. CRL-6475 in 2020). B16-F10 was transduced with MSCV γ-retroviral particles encoding a full-length ovalbumin (OVA) protein in tandem with a T2A sequence and a blasticidin resistance insert. B16-OVA-T2A blastidicin-resistant transductants were selected by supplementing culture medium with 10 μg/mL Blasticidin S HCl (Gibco). The reactivity of OT-1 CD8⁺ T cells against the selected B16-OVA-T2A tumor line was validated using a coculture. Briefly, B16-OVA cells were pretreated overnight with 50 μg/mL recombinant murine IFNγ (PeproTech). A total of 5×10⁵ parental B16 or engineered B16-OVA cells were coincubated with an equal number of OT-1 CD8⁺ T cells in complete RPMI-1640 (see below) supplemented with 100 IU/mL recombinant human IL2 (PeproTech) for 6 hours at 37°C 5% CO₂. All cells were then fixed and processed to assess cytokine production of OVA-reactive OT-1 T cells by flow (see “flow cytometry”). Cell lines were maintained in complete media DMEM (Gibco) completed with 10% FBS (VWR), 1% glutamine (Gibco), 1% non-essential amino acids (Gibco), 1% penicillin–streptomycin (Gibco), 500 ng/mL amphotericin B (Gibco) and 1X Plasmocin Prophylaxis (Invivogen). B16-OVA cells were maintained in 10 μg/mL blasticidin. All cell lines are tested for mycoplasma via a PCR-based assay approximately once a week when in use. Vaccinia virus expressing the OVA epitope was a generous gift from Jeffery Nolz (OHSU – MMI). vv-OVA was generated by homologous recombination, where DNA encoding an OVA epitope was inserted into the TK gene of the parental strain of the vaccinia virus (Western Reserve) and selected by lack of TK activity. The vv-OVA strain was maintained by propagation in BSC-40 cells.

Murine CD8⁺ T cells were isolated from the spleens of OT-1 Ly5.1 transgenic mice using the EasySep Mouse CD8⁺ T-cell Isolation Kit (Stemcell). Spleens were first dissociated into a single-cell suspensions by mashing them and passing them through a 40 μmol/L filter, lysing for 30 seconds at room temperature using 1 mL ACK lysis buffer (Gibco), and then the resultant splenocytes were resuspended into EasySep Buffer (Stemcell) as per the manufacturer's instructions. CD8⁺ T cells were isolated via negative selection using an EasyEights EasySep Magnet (Stemcell), which yielded an enrichment of 96%–100% purity. Human CD4⁺ and CD8⁺ HD T cells were purified from fresh whole-blood via gradient with Ficoll-Paque PLUS Media (Cytivia). Briefly, 2 mL of blood were diluted 1:1 with Ca²⁺- and Mg²⁺-free DBPS (Gibco), and then interfaced with 3 mL of ficoll media in a 50 mL conical. These quantities were scaled according to the total volume of the sample. The interfaced sample was then centrifuged for 30 minutes at room temperature with an acceleration of 3 and the brake turned off. Finally, lymphocytes were collected from the resultant plasma–ficoll interface, where this fraction contained 60% CD45⁺CD4⁺ or CD45⁺CD8⁺ T cells.

CD8⁺ T cells from OT-1 mouse spleens were stimulated in vitro with plate-bound anti-CD3 and anti-CD28 (5 μg/mL; BioXcell) for 24 hours and expanded in RPMI-1640 (Gibco) completed with 10% FBS (Gibco), 1% nonessential amino acids (Gibco), 1% sodium pyruvate (Gibco), 1% penicillin/streptomycin (Gibco), 57 μmol/L 2-BME (Sigma), and 100 IU/mL of recombinant human (rh)IL2 (Peprotech). Mouse T cells that underwent a second 24-hour stimulation 6 days after the first stimulation are identified in the text as “serially” TCR stimulated. Isolated human T cells were stimulated in vitro for 48 hours in 50:50 complete RPMI-1640 (Gibco): OpTmizer CTS (Gibco) containing a 1:100 titer of TransAct (Miltenyi Biotec), 5% FBS, and 300 IU/mL of rhIL2 (Peprotech). Following stimulation with TransAct, Human T cells were maintained in 50:50 complete RPMI-1640 (Gibco): OpTmizer CTS (Gibco) 300 IU/mL of rhIL2 (Peprotech). Human T cells were exposed to 50 nmol/L ouabain (Sigma) as indicated. For antioxidant treatment, cells were cultured with 10 mmol/L NAC (Sigma). For measuring T-cell effector function, cells were washed of their original culture media (experimental or control and cytokines) before stimulation for 5 hours with anti-CD3 and anti-CD28 or with 50 ng/mL PMA (Sigma) and 1.5 μmol/L ionomycin (Sigma), without IL2, and in the presence of brefeldin A and monensin (BD Biosciences). One of these 6-hour anti-CD3 and anti-CD28 stimulation treatments is identified in the text as an “acute” TCR stimulation.

Platinum-E ecotropic packaging cells (Cell Biolabs; Cat. RV-101) were plated one day before transfections on poly-D-lysine-coated 10-cm plates (Corning) at a concentration of a total of 6 × 10⁶ cells per plate. Packaging cells were transfected with 20 μg of plasmid DNA containing indicated genes of interest synthesized by Genscript, Inc. and cloned in an MSCV derived γ-retrovirus backbone derived from Addgene Plasmid #122727along with 3 μg pCL-Eco plasmid DNA (Addgene, Plasmid #12371), using 60 μL Lipofectamine 2000 in OptiMEM (Invitrogen) where indicated, for 8 hours in antibiotic-free complete DMEM. MSCV K_v1.3_DN was obtained by site directed mutagenesis performed by Genscript, Inc. of K_v1.3_WT (6) to mutate glycine residues 399 and 401 to alanines (GYG to AYA). MSCV K_ir2.1 was generated by codon-optimized (jcat.de) synthesis of the wild-type mus musculus amino acid sequence (P35561) and EcoRI-NcoI–based digestion and ligation into MSCV by Genscript, Inc. Medium was replaced 24 hours after transfection, and cells were incubated for a further 48 hours. Retroviral supernatants were then collected and centrifuged at 2,000 x g for 2 hours at 32°C in 24-well, non-tissue culture-treated plates coated overnight in Retronectin (Takara Bio) or spinfected for 90 minutes at 32°C in 1 μcg mL⁻¹ polybrene (Sigma), which was then removed after an additional 4-hour incubation.

Non-targeting control single-guide (sg)RNA and Atp1a1-targeted modified sgRNAs were purchased from and synthesized by Synthego. Modified sgRNAs were designed to target protein-coding loci of the Mus musculus Atp1a1 gene (Genbank #AC127357.5) and Homo sapiens ATP1a1 gene (Genbank #AH001423.2). Modified sgRNAs were mixed in equal parts (murine Atp1a1 sgRNAs: 5′-AAAAGCCUCCAAAGAGCUGC-3′, 5′-GAGGGGGUGUGAGGGCGUUG-3′, 5′-CUUGCAGGUGUUAAUCCCUG-3′; human ATP1A1 sgRNAs: 5′-UGCUCGUGCAGCUGAGAUCC-3′, 5′-CCAUUCAGGAGUAGUGGGAG-3′, 5′-UGGAGCGAUUCUUUGUUUCU-3′). Mixtures of complexed sgRNA and Cas9 (0.3 nmol synthetic sgRNA + 62 μmol Cas9 nuclease (Integrated DNA Technologies) and a total of 2–10×10⁶ enriched murine CD8⁺ T cells or human PBMCs following FICOLL density centrifugation were suspended in 25 μL of P3 electroporation buffer (Lonza) and electroporated using the Lonza 4D Nucleofector (pulse code DN100). Cells were activated ex vivo as above and collected for INDEL assessment on day 4 of culture.

For INDEL assessment, genomic (g)DNA was harvested from mouse and human cells at time of analysis using the DNeasy Blood and Tissue DNA kit (Qiagen), and the region containing the Atp1a1/ATP1A1 sgRNA target was PCR-amplified and Sanger sequenced by Eurofins Genomics. For murine samples, a 435-bp region was amplified for sequencing (forward primer: 5′-AGCAGAATGACCCAGAGTGG-3′, reverse: 5′-GGGGGAGTTAACCAGGCTTC-3′); for human samples, a 498-bp region was amplified (forward: 5′-GTGGGGACTGGCTCATCAG-3′, reverse: 5′-GAGTTCATAACCATTAAGTAATGAGTGG-3′). 100 to 200 ng of each amplicon was sent for sequencing with 2 pmol of its respective sequencing primer (mouse: 5′-GAGAAGAAGGGACAAATGTGAGCAG-3′; human: 5′-AGTTCATAACCATTAAGTAATGAGTGGTAA-3′). Genomic fragments were analyzed using a 3730xl DNA Analyzer (Applied Biosystems). The indel percentage of amplicons derived from Atp1a1-deficient cells was deduced using the Synthego ICE tool (Synthego Performance Analysis, ICE Analysis. 2019. v3.0. Synthego).

For all flow cytometry experiments, T cells were stained with a fixable live/dead stain (Invitrogen) in PBS, followed by surface antibody staining (see Supplementary Table 1 unless otherwise indicated) in FACS buffer (PBS with 0.5% BSA and 0.1% sodium azide, Sigma) for 30 minutes at 4°C. For intracellular cytokine staining, cells were first stained for surface markers and then stained for intracellular molecules overnight following fixation and permeabilization following the manufacturer's protocols (BD Cytofix/Cytoperm). All antibodies were diluted in 1:200 for staining, except for phosflow antibodies (pS6^S235/6 PE-Cy7, pAKT^S473 AF488, pERK^T202/Y204 Pacific Blue (Cell Signaling Technology), and pCD3^Y142 PE (BD Biosciences), which were diluted in 1:400 pCD3ζ^Y142 PE (BD Biosciences).

For phosflow, T cells were incubated with 5 μcg mL⁻¹ anti-CD3 (145–2C11) and anti–CD28-biotin (37.51) conjugates (eBioscience). TCR cross-linking was carried out at 37°C by the addition of soluble streptavidin (Thermo Fisher Scientific) to a final concentration of 20 μcg mL⁻¹. Reactions were quenched at specified time points by the addition of warmed lyse/fix PhosFlow Buffer I (BD Biosciences). Cells were then washed and permeabilized with –80°C PhosFlow Buffer III (BD Biosciences). After washing, cells were stained with phosphoantibodies (Supplementary Table 1) purchased from Cell Signaling Technology or BD Biosciences at a concentration of 1:400 for 1 hour. All experiments were conducted on a BD (Becton Dickinson) Fortessa or Cytek Biosciences Aurora cytometer and analyzed with FlowJo software (Tree Star, Inc.).

FACS purification of CD44⁻CD62L⁺ CD8⁺ T_N and CD44⁺CD62L⁻ T_EM cells from the spleens of donor Rag2^−/−CD45.2⁺ mice and CD45.1⁺ recipient mice was performed on a BD FACS Aria. In all instances, spleens were harvested and dissociated into a single-cell solution as previously described, then stained for viability with propidium iodide and for CD8, CD44, and CD62 L (Supplementary Table 1). Purification of CD45.2⁺ donor cells from CD45.1⁺ recipient mice additionally required staining of CD45.1 and CD45.2. Following a wash and resuspension in FACS buffer, live, CD19⁻, CD8⁺ T cells were gated and sorted into T_N (CD44⁻CD62L⁺) and/or T_EM (CD44⁺CD62L⁻) populations. These populations were additionally gated on CD45.2 during FACS purification of CD45.2⁺ OT-1 donor cells from CD45.1⁺ recipients. In both experiments, the T_N fraction had 99.9% purity upon validating the sorted fractions; the T_EM fraction had 98.6% purity.

Murine CD8⁺ T cells purified from spleens were rested in IL2-free complete RPMI-1640 (as defined above) for 16 to18 hours. Cells were stained with fixable live/dead and anti-CD8 (Supplementary Table 1), washed, and then resuspended with 5 μmol/L DiSBAC₂ (ref. 3; Invitrogen), 2.5 μmol/L DCFDA (AdipoGen Life Sciences), or 1 μmol/L APG-4 (Asante Green-4, TEFLabs) in complete RPMI-1640 and incubated for 30, 20, and 90 minutes at room temperature, respectively. Cells stained for oxonol were additionally stained for a Thy1.1 transduction marker, washed 1X and analyzed by flow cytometry. Only cells that were gated as live and CD8⁺ and, as necessary, Thy1.1⁺ were assessed for relative V_m. DCFDA- and APG-4–loaded cells were analyzed on a cytometer immediately after incubation without washing. For these assays, live, singlet CD8⁺ T cells were assessed for relative ROS abundance and intracellular [K⁺], respectively.

For determination of [K⁺]_i, we generated standard curves representing induced [K⁺]_I versus APG-4 mean fluorescence intensity (MFI) and used these standard curves to compute [K⁺]_i for each cell type. Briefly, standard curves accommodated [K⁺]_I ranging from 0 to 143 mmol/L, where these [K⁺]_i were induced by incubating cells with isotonic, [K⁺]-controlled RPMI-1640, 1 μmol/L APG-4 (pre-incubated 1:1 with Pluronic-127, Invitrogen), and K⁺-permeable ionophores valinomycin, gramicidin, and nigericin (Sigma) at 10 μmol/L, along with 50 mmol/L D-sucrose (Thermo Fisher Scientific) to normalize cell size. Cells queried for [K⁺]_I were incubated with APG-4 and isotonic, K⁺-free RPMI-1640 and live, CD8⁺ T cells were analyzed via flow cytometry. After flow analysis, we established linear regressions (r² ≥ 0.80) to deduce [K⁺]_I of the queried sample. Standard curves were generated for each unique cell type (i.e., Atp1a1-deficient cells, Atp1a1-sufficient cells).

For V_m and [K⁺]_i assessment, to control for technical variations in dye loading, a fluorometrically distinguishable loading control was included within each sample, typically a murine CD8⁺ T-cell retrovirally transduced to express Thy1.1 and stained CD90.1 PE-Cy7 (BD Biosciences). The quotient between the dye's MFI of experimental cells and loading control cells is depicted as relative MFI. The “Relative MFI” for both “Vehicle” and “K⁺ 20 mmol/L” was determined using the average of “Vehicle” replicates as the denominator.

Murine CD8⁺ T cells were isolated from spleens and primed as above. Before analysis, cells were rested in IL2-free complete RPMI-1640 for at least 8 hours. Cells were loaded with 1 μmol/L Fluo3-AM and 1 μmol/L Fura Red-AM (Invitrogen) for 30 minutes at 37°C in HBSS with 1.26 mmol/L Ca²⁺, 0.90 mmol/L Mg²⁺ and 2% FCS (Sigma), washed twice, and then resuspended in HBSS with 10 μg/mL anti-CD3 (145–2C11) and anti–CD28-biotin (37.51) conjugates (eBioscience), a live/dead stain and anti-CD8 (Supplementary Table 1). For flow cytometry analysis, samples were resuspended in 37°C additive-free RPMI, a baseline measurement was recorded for 20 seconds, followed by the addition of streptavidin (Invitrogen) to a final concentration of 20 μg mL⁻¹ to induce TCR cross-linking and Ca²⁺ influx. The BD Fortessa cytometer was used for the calcium influx assay. Kinetic analyses were performed with the FlowJo software package (TreeStar) and are gated on live CD8⁺ T cells.

We implanted C57BL/6 with subcutaneous B16-OVA melanoma (1 × 10⁶ cells). At the time of ACT, 10 days after tumor implantation, mice (n ≥ 5 for all groups) were sub-lethally irradiated (500 cGy, XStrahl CIX2), randomized, and injected intravenously with a total of 2 × 10⁶ OT-1 Ly5.1 cells electroporated with scrambled sgRNA or Atp1a1-sgRNA. A third group was injected intravenously with 100 μL of PBS as a nontreatment control. Mice used to assess tumor-infiltrating lymphocytes (TIL) 5 to 7 days after transfer were intravenously injected with a mixture containing a total of 6 × 10⁵ OT-1 (3 × 10⁵ of Scramble/Atp1a1 sgRNA treated each) cells total. Both cohorts received intraperitoneal injections of rhIL2 in PBS (5 × 10⁴ IU in 0.5 mL) once daily for 2 days starting on the day of cell transfer. Mice supplemented with NAC were additionally administered with intraperitoneal injections of NAC at a dosage of 0.25 mg/g once daily until the study's endpoint. NAC intraperitoneal injections were prepared in PBS, neutralized to pH = 7.2–7.8, and sterile-filtered before administration. For longitudinal tumor curve studies, tumors were measured using digital calipers every 2–3 days for 28 – 35 days. Size was measured in a blinded fashion approximately every two days after transfer, and tumor area was calculated as length × width of the tumor. Mice with tumors greater than 400 mm² were euthanized. The products of the perpendicular tumor diameters are presented as mean ± s.e.m. at the indicated times after ACT.

For administration of Rag2^−/−CD45.2⁺ OT-1 T_N and VV-OVA to CD45.1⁺, CD45.2⁺ T_N cells were immediately prepared for transfer after FACS purification at a dosage of a total of 5×10⁴ cells per CD45.1⁺ recipient. These donor cells were administered via intravenous injection.

B16-OVA tumors were analyzed for infiltrate of transferred cells to B16 tumors and spleens 5 to 7 days after ACT. Mice whose splenocytes and TILs were interrogated for in vivo IFNγ production were administered 100 μg Brefeldin A (Sigma) in PBS via retro-orbital injection 4 to 6 hours before euthanasia and tissue processing. Following euthanasia, tumors and spleens were promptly dissected, weighed, and stored in ice-cold PBS until further processing. Tumors were subsequently minced with surgical scissors, suspended into 3 mL of ice-cold PBS, dissociated into a single-cell solution using a GentleMACS octo-dissociator with heaters (Miltenyi Biotec, program = m_impTumor_03), filtered through a 100 μmol/L filter, and then consolidated and processed for flow cytometry (see above). Spleens were mashed and filtered through a 40 μmol/L filter, ACK lysed for 30 seconds using ACK lysing buffer (Gibco), and then consolidated and processed for analysis by flow cytometry. All B16-OVA tumor samples were stained with, at minimum, fixable live/dead, CD8, CD90.2, CD45.1, and CD45.2, along with phenotypic PD1 and CD62 L (Supplementary Table 1). B16-OVA tumors were analyzed on a Cytek Aurora spectral cytometer. Analyzed TILs are gated on a live, CD90.2⁺CD8⁺ population, and transferred populations are additionally congenically distinguished from endogenous CD90.2⁺CD8⁺ cells via CD45.1 and CD45.2.

Following flow cytometry and analysis, the relative abundance of co-transferred cells was corrected for all enumerative statistics (e.g., “%CD45.1⁺ of CD8⁺”) using the composition of the adoptively transferred cell mixture. Briefly, we assessed the input mixtures for relative abundance of each congenically marked, transferred cell type (i.e., CD45.1/2⁺ and CD45.1/1⁺ cells). These abundances were then used to assign a corrective constant to each component of the mixture, where the “major” component of the mixture was provided a constant of 1 and the “minor” component was assigned a constant of greater than 1. Following tumor harvest and flow cytometry, the observed abundances of TILs were corrected by multiplying these constants with the observed abundances. Statistics that represent a percentage of a whole (e.g., “%CD45.1/2⁺ of CD45.1⁺ cells”) were then further corrected to reflect both the corrected relative abundances of CD45.1/2⁺ and CD45.1/1⁺ cells at the time of transfer and the total component of the whole that both parts occupy.

RNA sequencing (RNA-seq) and determination of differentially expressed genes was performed by Azenta Life Sciences. Briefly, RNA was extracted from frozen murine CD8⁺ T-cell pellets using the RNeasy Plus Universal Mini Kit (Qiagen) as per the manufacturer's instructions. RNA integrity was checked using TapeStation (Agilent Technologies), and sequencing libraries were prepared using the NEBNext Ultra II RNA Library Prep for Illumina (New England Biolabs) following the manufacturer's instructions. Briefly, mRNAs were first enriched using Oligod(T) beadsFollowing mRNA fragmentation for 15 minutes at 94°C, first and second strand cDNA was synthesized using NEBNext Random Primers (New England Biolabs). cDNA fragments were end-repaired, adenylated at 3′ ends, ligated to NEBNext Adapters (New England Biolabs), and enriched by PCR. Samples were clustered onto a flowcell and sequenced on the Illumina hiSeq 4000 (Illumina) with a total of 2×150 bp paired end reads. Base-calling and raw sequence data were de-multiplexed using Illumina's CASAVA and bcl2fastq 2.20 software. Sequenced reads were trimmed using Trimmomatic v.0.36 (USADEL LAB, usadellab.org) and aligned to the mouse genome (ENSEMBL GRCm39) using the STAR aligner v.2.5.2b. Unique gene hit counts were deduced using the Subread package v.1.5.2, and only reads that fell within exon regions were retained for analysis. Differentially expressed genes were calculated using the R DESeq2 package. The Fisher's exact test or t test was used to evaluate significance with indicated P value and fold-change thresholds. Gene set enrichment analysis (GSEA) was carried out using GSEA v4.2.3 software with enrichment of the MSigDB hallmark gene set (12, 13).

CD8⁺ murine T cells or human lymphocytes were resuspended in 50–100 μL cell lysis buffer composed of 5 mL Mammalian Protein Extraction Reagent (Thermo Fisher Scientific), 50 μL Protease Inhibitor (Bimake), and 50 μL Phosphatase Inhibitors A and B (Bimake). Samples were then centrifuged at 20,000 × g for 10 minutes at 4°C to pellet cell debris. Proteins were extracted from the supernatant by adding NuPAGE LDS Sample Buffer (Thermo Fisher Scientific) supplemented with 5% 2-BME (Sigma) and heating the sample for 10 minutes at 75°C. Protein content was then quantified using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific), a fluorometric plate assay that used to generate an absorbance versus protein concentration standard curve and subsequently extrapolate protein concentration from extracts. Between 15 and 25 μg proteins were separated by 4%–12% SDS-PAGE, followed by standard immunoblot analysis using anti-Na⁺–K⁺-ATPase α1 (for mouse cells, Cell Signaling Technology #23565) at 1:1,000, anti-Na–K-ATPase (for human cells, Cell Signaling Technology #3010) at 1:1,000 or anti-GAPDH (as a loading control, Cell Signaling Technology) at 1:5,000 at 4°C overnight. Detection of proteins was performed using secondary antibodies conjugated to horseradish peroxidase, incubated at 1:5,000 for 1 hour at room temperature, and developed using the super signal west pico chemiluminescent substrate (Thermo Fisher Scientific). Blots were imaged on a GelDoc XR+ imager with Image Lab software (Bio-Rad).

Data were analyzed using an unpaired two-tailed Student t tests or two-way ANOVA. For adoptive transfer experiments, recipient mice were randomized before cell transfer. Tumor measurements were captured in a blinded fashion and plotted as the mean ± SEM for each data point. Tumor treatment graphs were compared by using the Wilcoxon rank sum test, and analysis of animal survival was assessed using a log-rank test. In all cases, two-tailed tests with P values less than 0.05 were considered significant. Statistics were calculated using GraphPad Prism 7 software (GraphPad Software Inc). Experimental sample sizes were chosen using power calculations, preliminary experiments, or were based on previous experience of variability in similar experiments. Samples that had undergone technical failure during processing were excluded from subsequent analysis.

RNA-seq data are available at GEO, accession numbers GSE242525 and GSE84996. Raw data for Fig. 1C were generated at the NIH Intramural Sequencing Center, and the derived data are available from the corresponding author upon request. Sequencing of genomic DNA for quantification of insertion and deletion frequencies can be accessed via Genbank Accession numbers OR795845-66.

We previously found that cancer cell death results in accumulation of K⁺ within the extracellular space of tumors, constraining mouse and human T-cell activation (6, 14). Here, we found that ↑[K⁺]_e, characteristic of tumors, suppressed cytokine production in response to TCR ligation in CD8⁺ T cells (Fig. 1A). In addition, ↑[K⁺]_e promoted T-cell suppression via elevation in [K⁺]_i, quantified by the fluorescence of the K⁺-sensitive dye APG-4, which could be reversed with the addition of the ionophore gramicidin (Fig. 1B). Gramicidin creates synthetic pores in the plasma membrane that are permeable to Na⁺ and K⁺, lowering [K⁺]_i, depolarizing V_m and restoring cytokine production in the presence of ↑[K⁺]_e (6).

To obtain a holistic understanding of K⁺ transporters in T cells, we FACS isolated naïve (T_N) CD44⁻CD62L⁺ CD8⁺ T cells from OT-I TCR transgenic Rag2^−/−CD45.2⁺ mice and transferred them to CD45.1⁺ hosts in combination with OVA-expressing vaccinia virus (VV-OVA). We then purified T_N (CD44⁻CD62L⁺) and T_EM (CD44⁺CD62L⁻) populations from the recipients and performed whole-transcriptomic analysis (Fig. 1C). Similarly, we activated CD8⁺ T cells for 2 hours via their TCR and assessed global transcriptional changes (Fig. 1D). We found high expression of transcripts encoding the channels K_v1.3 and K_Ca3.1, as expected. In addition, multiple other voltage gated, ligand gated, two-pore, and inwardly rectifying channels, along with alpha and beta subunits of the Na⁺/K⁺ ATPase, were differentially expressed across CD8⁺ T-cell subset and state. Drawing from these findings, we elected to interrogate the role of the Na⁺/K⁺ ATPase in CD8⁺ T cells, owing to its abundance and dynamism in CD8⁺ T cells, along with its importance in other cell types that maintain high intracellular [K⁺] (15).

The Na⁺/K⁺ ATPase is a heterodimeric enzyme consisting of catalytic (alpha) and regulatory (beta) subunits (15). Because both Atp1b1 and Atp1b3 isoforms of the β-subunit are present in CD8⁺ T cells, we targeted the nonredundant α isoform, encoded by the Atp1a1 locus, for genetic disruption (Fig. 1E and F). Using the fluorometric K⁺ dye APG-4, we found that CD8⁺ T cells lacking the Na⁺/K⁺ ATPase harbored an intracellular [K⁺]_i of 70.6±6 versus 130.8±9 mmol/L in control cells (Fig. 1G). In addition to maintaining [K⁺]_i, the Na⁺/K⁺ ATPase supports a negative resting V_m relative to the extracellular space (15). We used the voltage-sensitive membrane-permeable dye DiSBAC₂ (3), to comparatively assess membrane potential between CD8⁺ T-cell populations. As expected, T cells deficient for Atp1a1 maintained a depolarized V_m compared with scramble controls. CD8⁺ T cells transduced with retroviral particles encoding K⁺ channels that enforce a more negative membrane potential, that is, hyperpolarization (K_ir2.1 over expression), or increase the membrane potential closer to zero, depolarization (dominant negative K_v1.3), are included for comparison and assay validation (Fig. 2A). Concordantly, we found that Atp1a1 disruption compromised Ca²⁺ influx following both TCR- and ionomycin-induced store operated calcium entry (Fig. 2B and C). Yet, CD8⁺ T cells lacking the Na⁺/K⁺ ATPase had abundant phosphorylation within a number of signal transduction pathways. Most notably, CD3ζ sustained high levels of phosphorylation, along with Akt–mTOR and MAPK/Erk pathway members (Fig. 2D). Consistent with increased activation within multiple signaling cascades, Atp1a1-deficient CD8⁺ T cells harbored high expression of activation and exhaustion markers, such as PD-1, Tim-3, Fas, and CD69, with a relative loss of the stemness-associated proteins Tcf7 and CD62 L (Fig. 2E and F). Atp1a1-deficient T cells also exhibited diminished effector function (Fig. 2G; refs. 16–18).

Despite contributing to a resting negative V_m, and therefore promoting TCR-induced Ca²⁺ influx, the primarily role of the Na⁺/K⁺ ATPase in T cells is to constrain tonic signal transduction that otherwise disrupts T-cell quiescence and promotes T_Ex cell formation (19, 20). To better understand how the Na⁺/K⁺ ATPase controlled T-cell behavior, we first performed whole-transcriptomic analysis of C57BL/6 CD8⁺ T cells four days after CRISPR-Cas9–mediated disruption of Atp1a1. Transcriptionally, Atp1a1 disruption promoted effector T-cell programs, stress response pathways, and anabolic metabolism, with a coincident loss of stemness-associated programs (Fig. 3A and B). GSEA of expression profiles increased following Atp1a1 disruption identified signatures canonical for stress response, effector T-cell programs, including mTORC1 and ROS metabolism among others (Supplementary Fig. S1). Consistent with these transcriptional changes, Atp1a1-deficient T cells harbored elevated levels of ROS, quantified by the fluorometric dye DCFDA (Fig. 3C).

Recent observations that the antioxidant NAC attenuates ROS, limits terminal differentiation, and promotes stemness in T cells prompted us to apply NAC to Atp1a1-deficient T cells in an attempt to reverse their T_Ex cell phenotype (21). Antioxidant-mediated ROS neutralization prevented the accelerated differentiation otherwise observed following Atp1a1 deletion, evidenced by normalization of PD-1, Tim-3, and other markers associated with advanced differentiation, but did not restore cytokine production (Fig. 3C and D; Supplementary Fig. S2A and S2B).

Drawing from our prior observations (Fig. 1A and B; ref. 6), we hypothesized that provision of exogenous K⁺ could raise [K⁺]_i of Atp1a1-deficient T cells and prevent associated dysfunction. Indeed, ↑[K⁺]_e raised [K⁺]_i in Atp1a1-deficient T cells, limiting their activation and exhaustion (Fig. 3E; Supplementary Fig. S3A). Restoration of [K⁺]_i also limited ROS accumulation in Atp1a1-deficient T cells (Fig. 3F). These results indicated that [K⁺]_i may have a direct impact on ROS metabolism. Of note, both Scramble and Atp1a1-deficient CD8⁺ T cells exposed to NAC during their initial activation and expansion acquired a defect in cytokine production (Supplementary Fig. S2B), in line with prior results (21–24). These observations prompted us to interrogate whether [K⁺]_i might influence ROS metabolism following TCR ligation. Indeed, in both serial TCR stimulation and acute TCR stimulation, ↑[K⁺]_e limited exhaustion, ROS accumulation, and effector function of CD8⁺ T cells (Supplementary Fig. S3B and S3C). Taken together, these results indicate that [K⁺]_i influences ROS metabolism in genetically intact activated T cells, and that suppression of TCR-induced effector function and exhaustion by elevated intracellular K⁺ could be mediated through ROS metabolism.

We next asked whether these observations held true in human CD4⁺ and CD8⁺ T cells. We confirmed CRISPR-Cas9–mediated disruption of ATP1A1 at the protein level via immunoblot following electroporation with Cas9 complexed to ATP1A1 sgRNAs (Fig. 3G). Genetic disruption of the Na⁺/K⁺ ATPase promoted the retention of the activation marker CD69 and loss of CD62 L in both CD4⁺ and CD8⁺ T cells (Fig. 3H and I). We also found that treatment with the Na⁺/K⁺ ATPase inhibitor ouabain during activation and expansion resulted in acquisition of CD69 and loss of CD62 L in CD4⁺ and CD8⁺ T cells (Fig. 3J; refs. 25, 26). These findings indicate that low [K⁺]_i is the primary determinant of T-cell exhaustion in the setting of Na⁺/K⁺ ATPase disruption and suggest a link between [K⁺]_i and ROS metabolism. In aggregate, our results suggest that T-cell dysfunction induced by low [K⁺]_i is mediated, in part, by an accumulation of ROS that results in tonic signal transduction.

To assess the relevance of the Na⁺/K⁺ ATPase for T-cell function in vivo, we adoptively transferred congenically distinguishable OT-I TCR transgenic CD8⁺ T cells following electroporation with Cas9–sgRNA complexes targeting Atp1a1 or scramble sequences into C57BL/6 mice bearing established B16-OVA subcutaneous melanoma. Atp1a1-deletion severely compromised T-cell persistence after adoptive transfer within secondary lymphoid organs and tumors (Fig. 4A–C). Demonstrative of their defect in persistence, both tumor regression and host survival were significantly impaired in mice receiving OT-I CD8⁺ T cells lacking Atp1a1 (Fig. 4D and E). Phenotypic analysis of transferred populations demonstrated Atp1a1-deficient OT-I CD8⁺ T cells were in a more exhausted state within the tumor and spleen (Fig. 4F; Supplementary Fig. S4A–S4C).

To test whether ROS neutralization could prevent dysfunction and exhaustion of Atp1a1-deficient T cells in vivo, we again co-transferred CD45.1/2⁺ congenically distinguishable OT-I CD8⁺ T cells into CD45.2⁺ hosts. Despite ongoing exposure to NAC, Atp1a1-deficient cells exhibited a similar loss in persistence in vivo (Fig. 4G; Supplementary Fig. S4D and S4E and S4G and S4H). Ongoing antioxidant provision also failed to augment IFNγ production or rejuvenate the T_Ex cell state of Atp1a1-deficient CD8⁺ T cells (Fig. 4G–I; Supplementary Fig. S4D–S4I). Collectively, these results indicate that the Na^+/K⁺ ATPase plays a critical role in supporting T-cell homeostasis and function by maintaining [K⁺]_i and preventing tonic signal transduction.

The dynamic electrochemical gradients enacted by monovalent ions have been primarily studied for their role in supporting the function of “excitable” cells (i.e., expressing voltage–gated sodium and calcium channels) such as neurons, myocytes, and cardiomyocytes (27, 28). Prior work in T cells has focused on the importance of K_v1.3 and K_Ca3.1 in maintaining the V_m to facilitate TCR-induced Ca²⁺ influx (9, 29, 30). Others have documented the function of the Na⁺/K⁺ ATPase in peripheral T cells, speculating on the dynamics of its function across T-cell subsets and thymocytes (10). We previously reported that ↑[K⁺]_e requires PP2A activity to limit Akt–mTOR signaling and effector function in T cells, independent of V_m and Ca²⁺ (6). Here, we again found that the impact of [K⁺] upon T cells could not be accounted for by Ca²⁺ alone. In support of a Ca²⁺-independent role for K⁺ is the observation that myeloid cell response to danger signals (inflammasome formation) requires depletion of [K⁺]_i and is also suppressed by ↑[K⁺]_e (31). Our observation that restoration of [K⁺]_i normalized ROS levels and T-cell functional state underlines the fundamental, but unexplored, role that [K⁺]_I plays in T-cell function. Although there is a wealth of precedent linking ROS to post-translational modification of signaling machinery, the exact relationship between [K⁺]_i and ROS metabolism remains unclear and warrants additional investigation.

Historically, ROS are considered harmful metabolic by-products that result from mitochondrial activity. More recently, ROS have been appreciated to act as pleiotropic signaling co-factors that control both cellular homeostasis and response to external stimuli (32). Multiple reports have recognized that the influence of ROS in T cells extends to TCR-induced signal transduction, nutrient consumption, and stemness (21–23). Specifically, ROS-mediated oxidation of phosphatase cysteine residues and suppression of their activity has been observed in multiple phosphatases that are central to T-cell function, such as PTP1B, SHP2, PP2A, and PTEN (33). In concert with our prior observations, these findings provide a framework in which [K⁺]_I limits ROS abundance, thus maintaining phosphatase activity by preventing their oxidation (1). Taken together, these findings identify a previously unrecognized connection between [K⁺]_I, ROS homeostasis, and regulation of signal transduction in T cells that warrants further investigation from both a biological and translational perspective.

Although we set out to generate T cells resistant to the ↑[K⁺]_e within tumors, the ultimate consequence of a persistently low [K⁺]_I is compromised T-cell antitumor function. However, drawing from the data presented here, we now show that low [K⁺]_i itself can drive T_Ex cell formation. Thus, the next steps to engineer more effective tumor-specific T cells should include cell-intrinsic approaches to increase ROS neutralization (i.e., viral/ non-viral means to express ROS-neutralizing enzymes in a constitutive or regulable fashion), consideration of viral or non-viral means to augment Na⁺/K⁺ ATPase activity with constitutive or regulable expression of ROS producing enzymes (i.e., NAPDH oxidases for cytoplasmic ROS), and ex vivo expansion of mouse and human tumor-specific T cells in altered concentrations of [K⁺]_e to limit activation and exhaustion. Broadly, these findings advance a paradigm, where K⁺ transport is central to antitumor function and can be targeted to change T-cell behavior and improve cancer immunotherapies.

C. Collier reports grants from AACR, American Society of Clinical Oncology (ASCO), Pancreatic Cancer Action Network, Lyell Immunopharma, and NIH during the conduct of the study. M. McWhorter reports grants from AACR, PanCAN, Lyell Immunopharma, and NIH during the conduct of the study. C. Jenkins reports grants from AACR, PanCan, ASCO, Lyell, and NIH during the conduct of the study. R. Roychoudhuri reports personal fees from E3DG Genomics, as well as grants from AstraZeneca and F-Star Therapeutics outside the submitted work. R. Eil reports other support from Lyell Immunopharma, as well as grants from American Association of Cancer Research, ASCO, Pancreatic Cancer Action Network, and NIH/NCI (grant no. K08CA256179) during the conduct of the study; and a patent for US 2021/0332326 A1 issued and licensed to Lyell Immunopharma. No disclosures were reported by the other authors.

C. Collier: Conceptualization, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. K. Wucherer: Conceptualization, validation, investigation, methodology, writing–review and editing. M. McWhorter: Formal analysis, investigation, visualization, methodology, writing–review and editing. C. Jenkins: Validation, investigation, writing–review and editing. A. Bartlett: Conceptualization, formal analysis, investigation, methodology, writing–review and editing. R. Roychoudhuri: Conceptualization, methodology, writing–review and editing. R. Eil: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.