Nitric oxide (NO) is a signaling molecule produced by NO synthases (NOS1–3) to control processes such as neurotransmission, vascular permeability, and immune function. Although myeloid cell–derived NO has been shown to suppress T-cell responses, the role of NO synthesis in T cells themselves is not well understood. Here, we showed that significant amounts of NO were synthesized in human and murine CD8+ T cells following activation. Tumor growth was significantly accelerated in a T cell–specific, Nos2-null mouse model. Genetic deletion of Nos2 expression in murine T cells altered effector differentiation, reduced tumor infiltration, and inhibited recall responses and adoptive cell transfer function. These data show that endogenous NO production plays a critical role in T cell–mediated tumor immunity.
Nitric oxide (NO) is a membrane-permeable and highly labile gas with manifold effects on both cells and tissues. It has a short half-life in vivo, from seconds to a few minutes, and is rapidly oxidized to nitrate and nitrite (1). NO is known to play an essential role in a wide range of physiologic processes, including vascular permeability, neuronal function, and immune response (2).
Mammalian tissues can reduce nitrates and nitrites to bioactive nitrogen oxides, but the majority of NO is produced by the nitric oxide synthase (NOS) enzymes (3). In mammals, there are three NOS isoforms (NOS1–3) that synthesize NO and L-citrulline from L-arginine and oxygen. NOS1 and NOS3 are calcium-sensitive enzymes, predominantly expressed in neurons and in endothelial cells, respectively, whereas the inducible NOS2 is expressed most highly in immune cells of the myeloid lineage (4). Myeloid cell–derived NO has been shown to, amongst other things, allow the resolution of inflammation through suppressing T-cell responsiveness (5–8). However, the overall significance and function of endogenous NO synthesis by T cells is still not clearly defined (9–14).
T cells cultured in vitro in the presence of high NO concentrations, whether derived from myeloid cells or from NO-donor drugs, have reductions in proliferative responses and cytokine release (5, 15, 16). However, at low doses, NO can potentiate the expansion of T-cell populations and modulate T-cell fate and metabolism (17–20). This strongly suggests that the effect of NO on T-cell function is dose-dependent. Reports using activated T cells have identified expression of NOS1 and NOS3 in human cells, whereas murine T cells predominantly express NOS2 (17, 21–23). NO production by T cells has been shown to modulate the structure of the immune synapse and to regulate T-cell differentiation (11, 12, 14). This finding is somewhat controversial, however, because the expression of the NOS isoforms, as well as NO production, was not be detected in some studies of resting and activated T cells (9, 10, 13).
In this study, we investigated the role of endogenous NO production in CD8+ T cells and use Nos2 deletion models specific to T cells to determine the role of NO synthesis during tumor immunity. We found that endogenous NO production by T cells has a number of unexpected roles, including the facilitation of tumor infiltration. This has implications for both the natural function of movement of activated T cells through tissues, and for the modulation of immune function by pharmacologic means, including NO donors and NOS inhibitors.
Materials and Methods
Animal work was approved by the regional animal ethics committee of Northern Stockholm, Sweden.
C57BL/6J (CD45.2) animals were purchased from Janvier Labs. Donor T-cell receptor (TCR)–transgenic OT-I mice (JAX #003831; ref. 24) were crossed with mice bearing the CD45.1 congenic marker (JAX #002014; ref. 25). The targeted deletion of Nos2 was generated by our laboratory and created by crossing homozygous mice carrying loxP sites flanking exon 3 of Nos2 (26) into a mouse strain of Cre-recombinase expression driven by the distal promoter of the lymphocyte-specific Lck gene (JAX #012837; ref. 27). Targeted deletion of Hif1a was generated by our laboratory, and achieved by crossing homozygous mice carrying loxP sites flanking exon 2 of Hif1a (JAX #007561; ref. 28) into the distal promoter Lck cre transgenic strain described above. TdTomato reporter mice were generated in our laboratory by crossing homozygous mice carrying loxP-flanked STOP cassette associated to the tdTomato allele (JAX #007914; ref. 29) into the distal promoter Lck cre transgenic strain described above. All experiments were performed with age and sex-matched Cre-negative controls. Genotyping was performed by Transnetyx using real-time PCR. All mice used in the study were rederived from sperm into C57BL/6J recipients and used in a period of 2 years during which breeding pairs were selected to minimize inbreeding.
B16-F10 cells were originally purchased from ATCC (CRL-6475) and genetically modified to express chicken ovalbumin (OVA), eGFP, and neomycin phosphotransferase (30). The resulting OVA-expressing B16F10 cells were selected with 0.75 mg/mL G418 sulfate (ThermoFisher). HEK293 cells were a gift from Prof. Dantuma (Karolinska Institutet, Stockholm), and MC38 cells were a gift from Dr. Asis Palazon (University of Cambridge). Human umbilical vein endothelial cells (HUVEC) were obtained from ThermoFisher (C0035C). With the exception of HUVEC, cell lines were cultured in high-glucose DMEM with pyruvate (11995065, ThermoFisher) supplemented with 10% FBS (Sigma). HUVECs were cultured with EGM-2 Endothelial Cell Growth Medium (CC-3162, Lonza) containing EGM-2 SingleQuots Supplements (CC-4176, Lonza). All cells lines were cultured in the presence of 100 U/mL penicillin (Sigma) and 100 μg/mL streptomycin (Sigma) and in incubators with 5% CO2. Except for cells obtained directly from the supplier, cell lines were initially Mycoplasma tested using the MycoAlert Mycoplasma Detection Kit (LT07–118, Lonza). Cell lines were frozen at low passage number (<5) in DMEM containing 10% DMSO (Sigma) and were typically passaged 3 to 4 times between thawing and experimental use. Cell lines were not authenticated.
T-cell isolation and activation
Splenic murine CD8+ T lymphocytes obtained from C57Bl6/J mice were purified with either positive or negative selection using microbeads (130–117–044 or 130–104–075, Miltenyi Biotec) by magnetic-activated cell sorting (MACS) and cultured in T-cell media: RPMI (21875, ThermoFisher), 55 μmol/L β-mercaptoethanol (Sigma), 10% FBS, 100 U/mL penicillin, and 100 μg/mL streptomycin. Polyclonal mouse CD8+ T cells were activated with anti-mouse CD3/CD28 Dynabeads (ThermoFisher) at a 1:1 cell-to-bead ratio. Purified splenic OT-I CD8+ T cells were activated with 0.1 to 1 μg/mL OVA-derived peptide SIINFEKL (ProImmune) or with anti-mouse CD3/CD28 Dynabeads at a 1:1 cell-to-bead ratio. Mouse CD8+ T cells were expanded for up to 5 days in the presence of 10 U/mL recombinant human IL2 (Sigma). Human CD8+ T cells were purified from donor peripheral blood mononuclear cells (NHSBT or Karolinska Hospital) by positive CD8 MACS (130–045–201, Miltenyi Biotec), cultured for up to 5 days in T-cell media (without β-mercaptoethanol) supplemented with 30 U/mL recombinant human IL2 and activated with anti-human CD3/CD28 Dynabeads (ThermoFisher) at a 1:1 cell-to-bead ratio.
Splenic murine naïve CD4+ T lymphocytes were purified from C57Bl6/J mice with negative selection using microbeads (130–104–453, Miltenyi Biotec) by MACS and cultured in T-cell media: RPMI (21875, ThermoFisher), 55 μmol/L β-mercaptoethanol, 10% FBS, 100 U/mL penicillin, and 0.1 mg/mL streptomycin. For Th1 polarization, cells where cultured for 3 days with 20 ng/mL IL12 (R&D Systems) and 10 μg/mL anti-mouse IL4 (BioXCell). For regulatory T-cell (Treg) polarization, cells were cultured for 3 days with 2 ng/mL TGFβ (R&D Systems). For Th17 polarization, cells were cultured for 3 to 5 days with 2 ng/mL TGFβ, 20 ng/mL IL1β (R&D Systems), 20 ng/mL IL6 (R&D Systems), 10 μg/mL anti-mouse IL4, and 10 μg/mL anti-mouse IFNγ (BioXCell). CD4+ T cells were activated with anti-mouse CD3/CD28 Dynabeads (ThermoFisher) at a 1:1 cell-to-bead ratio and cultured for 2 to 5 days before analysis. T cells were cultured at a density of approximately 5 × 105 to 10 × 105 cells per mL per cm2. T-cell purity following MACS was confirmed to be greater than 95%.
Drug treatment of T cells
For each treatment, the same batch of cells was either incubated for 3 days with IL2 supplemented T-cell media containing the experimental compound or DMSO vehicle control (<0.1% of total well volume to avoid unspecific toxicity). The prolyl hydroxylase inhibitor FG-4592 (Cayman Chemicals) was used at 50 μmol/L in wild-type (WT) mouse CD8+ T cells cultured in 1% O2. The NOS2 inhibitor 1400W dihydrochloride (Cayman Chemicals) was used at 100 μmol/L in mouse CD8+ T cells transduced with VC or NOS2OE vectors following enrichment of Thy-1.1+ cells using MACS as described below. Human CD8+ T cells were treated with the NO donor compound 2,2′-(Hydroxynitrosohydrazino) bis-ethanamine (NOC-18; Cayman Chemicals) or with the panNOS inhibitor N(gamma)-nitro-L-arginine methyl ester (L-NAME; Cayman Chemicals) at concentrations ranging from 1 to 256 μmol/L. T cells treated with the different compounds were then analyzed as detailed in figure legends.
Culture media conditioned by CD8+ T cells for 1 day (human) or 3 days (mouse) was harvested and kept on ice. Human CD8+ T cells cultured for 3 days were lysed with 100 μL RIPA buffer (Thermo), and analysis of supernatants was performed following centrifugation at maximum speed for 15 minutes. Nitrite levels were quantified using a Sievers Nitric Oxide Analyzer (NOA 280i) according to the manufacturer's instructions. Media incubated without cells or lysis buffer was used to blank for basal nitrite signals. When cell number was not the same across all conditions, results were normalized to cell counts determined by a TC20 automated cell counter (Bio-Rad) or with the use of counting beads (C36950, ThermoFisher) followed by analysis with an Aurora flow cytometer (Cytek Biosciences).
Total RNA was extracted from isolated CD8+ T cells (RNeasy Kit, Qiagen), and 300 ng of RNA were used for cDNA synthesis (First-Strand Synthesis kit, Invitrogen). qRT-PCR was performed with SYBR green (Roche) in a StepOnePlus system (Applied Biosystems). All kits were used according to the manufacturer's instructions. Samples were run in technical duplicates. The program GeNorm (31) selected Hprt as the most reliable housekeeping gene to be used in the study. Primers were designed with NCBI primer blast (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) and are listed in Supplementary Table S1.
Cell pellets were incubated with urea-tris buffer (8 mol/L urea, 50 mmol/L Tris-HCl (pH = 7.5), 150 mmol/L β-mercaptoethanol), sonicated twice for 45 seconds intercalated with 1-minute incubation on ice, and then centrifuged at 14,000 × g, 4°C for 15 minutes. Proteins were separated by SDS-PAGE and transferred to nitrocellulose (mouse proteins) or polyvinylidene difluoride (human proteins) membranes with a Trans-Blot Turbo Transfer System (Bio-Rad). Membranes were incubated for 1 hour at room temperature with a blocking solution (1X ROTI Block (A151.4, Roth) used as blocking solution for mouse proteins and 5% nonfat milk in PBS + 0.1% Tween-20 was used as blocking solution for detection of human proteins). Then, membranes were incubated in 50 mL tubes with a 3 mL of blocking solution containing the antibodies. Primary antibodies were used at a 1:1,000 dilution and incubated overnight at 4°C in a tube roller and secondary antibodies were used at a 1:5,000 dilution and incubated for 2 hours at room temperature in a tube roller. After each antibody probing, membranes were washed 3×10 minutes in the 50 mL tubes with 10 mL of blocking solution. Mouse protein extracts (25–40 μg) were probed with primary antibodies against panNOS (CST #2977), NOS2 (CST #13120), and hypoxia-inducible factor (HIF)-1α (Novus NB-100–449) and detected using infrared-labeled donkey anti-rabbit (926–32213, LI-COR) and goat anti-mouse (926–68070, LI-COR) secondary antibodies in an Odyssey imaging system (LI-COR). The Revert 700 Total Protein Stain (TPS, 926–11011, LICOR) was used for the normalization of mouse protein expression. Human protein extracts (25–60 μg) were probed with antibodies against panNOS, NOS3 (CST #13120), and PPIB (CST #43603, for normalization) and detected using horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (HAF008, R&D Systems). Human proteins were detected with ECL Prime (GE Healthcare) and imaged with an iBrightCL1000 (ThermoFisher). M1-polarised mouse bone marrow–derived macrophage (BMDM) prepared as described below were used as positive control for mouse NOS2 expression whereas HUVEC cells were used as positive controls for human NOS3 expression.
In vitro cytotoxicity assay
10,000 B16-F10-OVA cells (target) were seeded per well in 96-well plates (flat bottom, Costar) and cocultured for a minimum of 14 hours with 2.5 × 103 to 2.5 ×105 mouse CD8+ OT-I cells previously activated for 3 days with 1,000 ng/mL SIINFEKL at 21%, 5%, or 1% O2. Wells were washed twice with PBS to remove T cells, and the number of remaining target cells was determined with Alamar Blue assay by culturing with 10 μg/mL resazurin (prepared from 4 mg/mL stock Sigma) and measuring the fluorescence signal (F, Ex/Em 530–560/590 nm) in a FLUOstar Omega plate reader (BMG Labtech). Cytotoxicity was calculated relative to wells with no T cells added (Pos CT) and wells with no B16 or T cells added (Neg CT): %Cytotoxicity = 100× [(Fsample – FNeg CT)/(FPos CT – FNeg CT)].
Mouse CD8+ T cells activated for 3 days in 1% O2 were assayed in a Seahorse Extracellular Flux Analyzer XF96 (Agilent) to determine oxygen consumption rate (OCR) and extracellular acidification rate (ECAR). The Seahorse assay was conducted in a hypoxia chamber at 3% O2. 1.5×105 CD8+ T cells were plated onto poly-D-lysine-coated wells in XF RPMI medium (Agilent), pH 7.4, supplemented with 10 mmol/L glucose (ThermoFisher) and 2 mmol/L glutamine (ThermoFisher). Media was preincubated at 1% O2. A minimum of 5 technical replicates per biological replicate were used. During the assay, wells were sequentially injected with anti-CD3/CD28 Dynabeads (4:1 bead to T-cell ratio), 1 μmol/L oligomycin (Sigma), 1.5 μmol/L FCCP (Sigma), and 100 nmol/L rotenone (Sigma) + 1 μmol/L antimycin A (Sigma).
Transwell migration assay
Mouse CD8+ T cells were purified from C57BL/6j mouse spleens with CD8a positive selection microbeads (Miltenyi Biotec) and cultured for 6 days in ambient oxygen before being transferred to 1% O2 for 48 hours. Mouse T-cell media was supplemented with 50 U/mL recombinant human IL2. Mouse primary lung endothelial cells (LEC) were isolated as previously described (32). LECs were cultured at 5% O2 in 6-well plates coated with 50 μg/mL collagen (Thermo) and cultured with EGM-2 Endothelial Cell Growth Medium containing EGM-2 SingleQuots Supplements and 100 U/mL penicillin and 100 μg/mL streptomycin. Two days before the assay, LECs were gently detached with 0.05% trypsin solution and 5×103 to 6×103 cells, plated in HTS Transwell-24 units with 3.0-μm pore size (Corning) coated with 50 μg/mL, and cultured at 1% O2. On the day of the assay, CD8+ T cells were loaded for 20 minutes at 37°C in PBS with 1 μmol/L calcein-AM solution (diluted in PBS from 1 mmol/L stock solution prepared by adding 50 μL DMSO to 50 μg lyophilized Calcein-AM from BioLegend). After washing and aspirating LEC media on transwells, 0.2×106 to 1×106 CD8+ T cells were added in a volume of 300 μL to the upper chamber, while the lower chamber contained 50 ng/mL murine CCL19 and murine CCL21 (R&D Systems). The assay was performed at 1% O2. The assay media was PBS containing 2 g/L glucose [used stock solution (ThermoFisher) at 200 g/L] and 200 mg/L arginine-HCl [used 10 g/L solution prepared from powdered arginine-HCl (Sigma) and pH adjusted to 7.4]. The calcein fluorescent signal (Ex/Em 494/517 nm) corresponding to T cells migrating through a LEC barrier was measured 3 hours after the start of the coculture in a FLUOstar Omega Microplate Reader (BMG Labtech). Transendothelial migration was calculated relative to wells with T cells added to the bottom chamber (MaxSignal) and wells with no T cells added (Neg CT): %Transendothelial migration = 100x [(Fsample – FNeg CT)/(FMax signal – FNeg CT)].
Analysis of lymphocyte populations in tissues
Thymus, inguinal lymph nodes, spleen, and peripheral blood were harvested from 8 to 15 weeks old female C57BL/6j Nos2fl/fl (control) and Nos2fl/fldlckCre animals. Thymus, spleens, and lymph nodes were mashed directly against 40-μm strainers using a syringe plunger and PBS + 0.5% FBS. After spinning at 500 g for 5 minutes, cell pellets were resuspended in 1 mL ACK Lysing Buffer (A1049201, ThermoFisher) and stained for flow cytometry analysis as described below. Blood was collected from the tail vein onto heparin-treated Capillary tubes (16.443, Sarstedt) and directly stained. Blood samples were treated with BD FACS lysing solution (BD 349202) prior to flow cytometry analysis.
Orthotopic tumor growth and tumor infiltration experiments
8- to 15-week-old female C57BL/6j mice were inoculated subcutaneously with 5×105 B16-F10-OVA or MC38 cells. For the orthotopic tumor growth experiment, tumor cells were inoculated in Nos2fl/fl (control) and Nos2fl/fldlckCre animals, and tumor volume was measured every 2 to 3 days with electronic calipers until day 30. Peripheral blood and tumor-infiltrating immune cell composition in Nos2fl/fl and Nos2fl/fldlckCre was analyzed 10 days after tumor inoculation. To assess infiltration of adoptively transferred OT-I CD8+ T cells in tumor-bearing animals, mice were conditioned for 11 days following tumor inoculation with intraperitoneal injection of 6 mg cyclophosphamide (CPA; approximately 300 mg/kg; Sigma). On day 14, 0.5×106 to 1×106 Thy-1.1-enriched transduced OT-I CD8+ T cells were intraperitoneally injected, and tissue infiltration was analyzed on day 19. Animals were assigned randomly to each experimental group. Tumors were processed in gentleMACS C tubes (Miltenyi Biotec). Tumors were minced in 4.5-mL HBSS (ThermoFisher) using sharp dissection scissors and processed in a GentleMACS dissociator (130–093–235, Myltenyi Biotec) using the mImpTumor-02 program. After adding 0.5 mL of HBSS solution containing 20% FBS, 10 mg/mL Collagenase Type IV (17104–019, Life Technologies) and 200 U/mL DNAse I (D5025, Sigma), tumor suspensions were incubated at 37°C for 1 hour under shaking. After further processing in a GentleMACS dissociator using the mImpTumor-03 program, tumor cell suspensions were transferred to 50-mL tubes through a 40-μm strainer. Cell suspensions from thymus, spleens, and liver were obtained by directly mashing the organs against 40-μm strainers using a syringe plunger and PBS + 0.5% FBS. After spinning at 500 g for 5 minutes, cell suspensions were resuspended in 1 mL ACK Lysing Buffer and stained for flow cytometry analysis as described below. As previously described, the blood was harvested from the tail vein and directly stained with fluorochrome-labeled antibodies and analyzed by flow cytometry after treatment with treated with BD FACS lysing solution.
Adoptive cell transfer of OT-I CD8+ T cells
Eight- to 15-week-old female C57BL/6j CD45.2+ mice were inoculated subcutaneously with 0.5×106 B16-F10-OVA and conditioned 4 days later with a peritoneal injection of 6-mg CPA per animal (approximately 300 mg/kg). On day 7, 0.5×106 to 1×106 CD45.1+ OT-I CD8+ T cells activated for 4 days with 100 ng/mL SIINFEKL were peritoneally injected. Animals were assigned randomly to each experimental group. Tumor volume was measured every 2 to 3 days with electronic calipers until day 50 to 60. Control and experimental groups are indicated in figure legends. Peripheral blood was collected from the tail vein at days 7 and 14 days after T-cell transfer and analyzed by flow cytometry as indicated below. Tumor volume was calculated using the formula a×b×b/2 where a is the length and b is the width of the tumor.
Generation of BMDMs
BMDMs were prepared from bone marrow extracted from the tibia and femur of WT C57BL/6J mice and cultured in non–tissue culture–treated Petri dishes in high glucose DMEM medium containing 10% FBS, 100 U/mL penicillin, 100 μg/mL streptomycin, and supplemented with 10 ng/mL mouse GM-CSF and M-CSF (R&D Systems). After 7 days of culture, BMDMs were activated with 100 U/mL EB lipopolysaccharide (LPS; Invivogen) for 24 hours (M1 polarization). After washing the monolayer, cells were detached after 10 minutes incubation in Cellstripper (Corning) using cell lifters. BMDMs in suspension were loaded with 100 ng/mL SIINFEKL peptide at 37°C for 1 hour and washed before being counted and injected in 1X PBS into mice as indicated below.
In vivo activation and recall experiment
OT-I T cells purified with CD8+ microbeads (Miltenyi Biotec) from spleens of Nos2fl/fl (WT; CD45.2) and Nos2fl/fldlckCRE (NOS2KO; CD45.1/CD45.2) C57BL/6j animals. WT (control) and NOS2KO cells were then mixed 1:1, and a total of 2 million cells (1 million of each genotype) was injected intraperitoneally into C57BL/6J CD45.1+ WT host mice. Endogenous and adoptive populations were distinguished by the allelic variants of CD45. One day later, host mice were vaccinated intraperitoneally with 8×105 SIINFEKL-loaded BMDMs. Peripheral blood was collected from the tail vein at days 7 and 10 after T-cell transfer and analyzed by flow cytometry following antibody staining and treatment with BD FACS lysing solution, as specified below. On day 30, animals were reinjected with 8×105 SIINFEKL-loaded BMDMs, and 7 days later, the spleens, inguinal lymph nodes, and liver tissues were harvested, mashed in 40-μm cell strainers into single-cell suspensions as previously described, and analyzed by flow cytometry to determine recall responses. Animals injected with PBS on day 30 were used as negative controls of the recall response. Animals were assigned randomly to each experimental group. Absolute numbers of adoptively transferred CD8+ OT-I cells were determined with the use of counting beads (CountBright, Life Technologies).
Single-cell suspensions were stained with Near-IR Dead Cell Stain Kit (ThermoFisher), followed by surface and intracellular staining in 96-well plates with 50 μL PBS + 0.5% FBS solution containing fluorochrome-labeled antibodies (Supplementary Table S2). Staining of cytoplasmic and nuclear antigens was performed using the Fixation/Permeabilization Kit (BD Biosciences) and the Transcription Factor buffer set (BD Biosciences), respectively. After each staining step, cell suspensions were washed twice with 200 μL of the buffer used to prepare antibody solutions and were centrifuged at 500 g for 2 minutes. To measure IFNγ, TNFα, and IL17 production, before intracellular staining, T cells were incubated in RPMI supplemented with 50 ng/mL phorbol 12-myristate 13-acetate (PMA; Sigma), 1 μg/mL ionomycin (Sigma), and 5 μg/mL Brefeldin A (Sigma) for 3 to 4 hours. When using blood and organ-derived cell suspensions, samples were treated with Mouse Fc Block (BD Biosciences) prior to antibody staining. For proliferation assays, nonactivated human CD8+ T cells and mouse CD8+ and CD4+ T cells isolated as shown above were loaded with CellTrace Violet (ThermoFisher) according to the manufacturer's instructions. Samples were acquired in FACSCanto II (BD Biosciences) or in Aurora (Cytek Biosciences) flow cytometers, and data were analyzed with FlowJo version 10.
Checkpoint blockade therapy
5×105 B16-F10-OVA cells were subcutaneously implanted in Nos2fl/fl (WT) and Nos2fl/fldLckCre (NOS2KO) animals. Ten days later, animals were administered 200 μg anti–programmed cell death protein 1 (PD-1) and anti-CTLA4 antibodies (InVivoMAb) or 200 μg isotype controls Rat IgG2a and Polyclonal Syrian Hamster IgG (InVivoMAb). Tumor volume was measured every 2 to 3 days with electronic calipers until day 45.
DNA encoding a codon-optimized polycistronic peptide composed of mouse Thy-1.1 and mouse NOS2 interspersed with picornavirus P2A and furin cleavage sequences were synthesized by GeneScript. The coding sequences were cloned into the gamma retroviral vector pMP71, a gift from Christopher Baum (MHH, Hannover). Addition of furin and self-cleaving picornavirus 2A sites enables posttranslational separation of Thy-1.1 and NOS2, whereas the polycistronic nature of the constructs ensures equimolar production of both proteins. A plasmid encoding Thy-1.1 alone was used as vector control (VC). Protein sequences and accession numbers are available in Supplementary Table S3. The use of plasmids encoding HIF constructs and their respective sequences have been previously described (30).
For the generation of retroviral particles, sub-confluent HEK293 cultures were transfected with NOS2 overexpression vector and pCL-Eco. Helper vector pCL-Eco (for ecotropic infection) was a gift from Inder Verma (Addgene plasmid \#12371). Supernatant media containing viral particles was harvested 48 hours after transfection and used fresh or stored at −80°C. Viral supernatants were spun onto Retronectin-coated wells (Takara) at 2,000 × g for 2 hours at 32°C and replaced with activated CD8+ T cells in T-cell media supplemented with 10 U/mL recombinant human IL2, with approximately 0.5 mL viral supernatant plated with 2×105 T cells per cm2. Three days after, cells were harvested, washed in a solution of PBS + 0.5% FBS and stained with Thy-1.1 (CD90.1) microbeads (130–121–273, Miltenyi Biotec) for enrichment of transduced cells with MACS following the protocol from the manufacturer.
Statistical analyzes were performed with Prism 9 software (GraphPad). A P value of < 0.05 was considered significant, and the statistical tests used are stated in figure legends.
The data generated in this study are available upon request from the corresponding author.
T cells express NOSs and produce NO
To understand the relevance of NO in T-cell biology, we activated CD8+ T cells isolated from mouse spleens in vitro and measured the expression of NOSs and the production of NO during the activation process (Fig. 1A). Because NOS expression has been shown to increase with low oxygen tensions in a HIF-dependent manner (33), we cultured cells in 21%, 5%, and 1% O2 for 1 to 3 days to assay differential NO production during activation at physiologically relevant oxygenation. Activated CD8+ T cells produced NO, as determined by extracellular nitrites (a byproduct of NO), with higher concentrations found in 1% O2 cultures (Fig. 1B; Supplementary Fig. S1A and S1B). Elevated NO correlated with increased NOS protein levels in T cells cultured in reduced oxygen (Fig. 1C). Nos2 mRNA expression increased following activation of T cells, and was more robust when the T cells were cultured in lower oxygen tensions (Fig. 1D). Activation of OT-I TCR transgenic CD8+ T cells with increasing doses of the cognate peptide SIINFEKL revealed a dose-dependent increase in expression of NOS2 (Fig. 1E). Enhancing HIF activity with the prolyl hydroxylase inhibitor FG-4592 further increased NOS2 protein expression (Fig. 1E), indicating that pharmacologic induction of the HIF pathway can directly augment NOS2 expression in T cells. Expression of NOS2 protein was reduced in T cells lacking HIF-1α (Supplementary Fig. S1C). Overexpression of HIF-1α, driven by various HIF constructs transduced into T cells (30), resulted in increased NOS2 protein expression; this increase correlated with increased HIF protein levels in the transduced cells (Supplementary Fig. S1D).
NO production was also detected in activated human CD8+ T cells and was correlated with increased Nos3 isoform mRNA and NOS3 protein levels (Fig. 1F–H; Supplementary Fig. S1E and S1F). These data show that NO was produced endogenously by both murine and human CD8+ T lymphocytes upon activation, despite the different NOS isoforms exhibited in each species.
Pharmacologic inhibition of NOS alters differentiation of human CD8+ T cells
To assess the role of NO in human T cells, we activated human CD8+ T cells in the presence of NOS inhibitor L-NAME or NO donor NOC-18 (Fig. 2A). Although human CD8+ T cells predominantly expressed NOS3, the use of a panNOS inhibitor, such as L-NAME, ensured maximal repression of NOS activity and was possible given its low toxicity at high concentrations ranging to 250 μmol/L (Fig. 2B; Supplementary Fig. S2A). We also used a NO donor compound to assess human T-cell growth and differentiation after exposure to high NO concentrations. NOC-18 was selected over other NO donors based on the slower kinetics of NO release and was detrimental to T-cell growth over the 3-day T-cell culture in the highest doses used (Fig. 2B and C; Supplementary Fig. S2A and S2B). L-NAME inhibited, in a dose-dependent manner, CD8+ T-cell expression of homing markers CD45RO, CCR7, and CD62L, as well as the key transcription factor T-bet (Fig. 2D; Supplementary Fig. S2C–S2E). NOC-18 only significantly altered CD62L expression (Supplementary Fig. S2E).
We confirmed that treatment with L-NAME reduced endogenous production of NO in human CD8+ T cells (Fig. 2E). Although the downstream target of TCR stimulation, phosphorylated S6, was not altered by L-NAME treatment (Fig. 2F), several T-cell activation markers, such as the IL2 receptor subunit CD25 and the cytotoxic marker CD107a (LAMP-1), were reduced after pharmacologic inhibition of NO production (Fig. 2G). This indicated that although exogenous high levels are inhibitory for T cells, endogenous NO production can modulate T-cell activation.
Mouse CD8+ T cells lacking NOS2 show altered effector differentiation and transendothelial migration capacity
To further characterize the role of NO production in T cells, we generated a mouse model with a T cell–specific deletion of Nos2. This was achieved by crossing homozygous mice carrying loxP sites flanking exon 3 of Nos2 (26) into a mouse strain of Cre recombinase expression driven by the distal promoter of the lymphocyte-specific Lck (dLck) gene (27). First, CD8+ T cells were isolated from Nos2fl/fldLckCre (NOS2KO) or Nos2fl/fl (WT) control animals, activated under various oxygen tensions for 3 days, and phenotypically characterized in vitro (Fig. 3). Upon activation, NOS2KO CD8+ T cells were unable to induce detectable expression of NOS2 protein (Fig. 3A) and produce NO (Fig. 3B). Despite not showing differences in expansion or proliferation following activation relative to WT control cells (Fig. 3C; Supplementary Fig. S3A), NOS2KO CD8+ T cells cultured in 1% O2 showed a significantly reduced proportion of terminally differentiated cells (CD44+CD62L−; Fig. 3D), as well as decreased expression of markers of T-cell activation (Fig. 3E). These included CD25, CD44 (cell adhesion receptor), ICOS, and CD27 (costimulatory molecule), granzyme B and IFNγ (effector molecules), and T-bet (key transcription factor for effector T cells). We also assessed the effect of Nos2 deletion in CD4+ T lymphocytes polarized in different oxygen tensions to generate Th1, Th17, and T regulatory cells (Supplementary Fig. S3B). With the exception of the increased proportion of IL17+ cells in Th17-polarizing conditions under 1% O2 in NOS2KO cells, no differences were found between NOS2KO and WT CD4+ T cells in terms of polarization or expression of differentiation markers.
Reactivation of CD8+T cells in 1%O2 showed that deletion of Nos2 reduced the increase in glycolytic rate that followed TCR triggering (as assessed by the ECAR; Fig. 3F). Nos2 deletion did not affect OCR or in vitro killing of tumor cells (Fig. 3F and G). Given the known role of NO in eliciting endothelium permeability, we used a coculture system to determine the degree to which T-cell infiltration through endothelial cell layers was dependent on endogenous T-cell NO production. In this experiment, T cells were inserted into a Boyden chamber above a layer of primary murine LECs. The cytokines CCL19 and CCL21 were placed in the lower compartment as chemoattractants. Absence of Nos2 expression significantly retarded transmigration of T cells through the endothelial cell layer (Fig. 3H).
To further understand the response of T cells to endogenous NO production, we also characterized CD8+ T cells overexpressing Nos2 (NOS2OE). We engineered a retroviral vector encoding a polycistronic peptide composed of Thy-1.1 (surface transduction marker) and mouse NOS2 (Supplementary Fig. S3C). CD8+ T-cell transduction with the NOS2OE vector increased NOS2 protein levels in Thy1.1+ cells when compared with VC or non-transduced (Thy1.1−) cells (Supplementary Fig. S3D). NOS2OE T cells enriched with Thy-1.1 beads by MACS (Supplementary Fig. S3E) increased NO production by 30-fold (Supplementary Fig. S3F). As expected with high NO levels, expansion of CD8+ T cells overexpressing Nos2 was severely impaired compared with controls, an effect that was abrogated in the presence of the NOS2 inhibitor 1400W (Supplementary Fig. S3G). In contrast to cells lacking Nos2, CD44 expression was increased in NOS2OE T cells relative to VC; this effect was also prevented by 1400W (Supplementary Fig. S3D and S3G). Nos2 overexpression also negatively impacted in vitro cytotoxicity (Supplementary Fig. S3H).
Nos2 deletion in T cells causes increased tumor growth
The specificity of dLck promoter activity in the T-cell compartment was confirmed in a Cre-loxP tdTomato reporter model (Supplementary Fig. S4A). Immune population frequencies were not altered in organs harvested from unchallenged NOS2KO animals compared with (WT) littermate controls (Supplementary Fig. S4B). We directly challenged NOS2KO animals with tumors to evaluate the effects of their specific deletion of Nos2 in the entire T-cell compartment (encompassing CD8+ and CD4+ T cells) on tumor growth (Fig. 4A).
NOS2KO animals bearing either MC38 or B16-F10-OVA cell line–derived tumors showed increased tumor growth, and a decreased survival rate was observed in both tumor models (Fig. 4B). Blood sampling on day 10 after tumor injection showed that NOS2KO animals bearing B16 tumors had a reduced proportion of terminally differentiated CD8+ T cells (CD62L−CD44+) compared with WT controls (Fig. 4C). Cell suspensions obtained from B16 tumors growing for 10 days revealed that the proportion of total and CD44+ subsets in both CD8+ and CD4+ T cells were reduced in NOS2KO animals compared with those subsets in tumors from WT control mice (Fig. 4D). The ratio between CD8+GZMB+ and CD4+FOXP3+ cells, as well as the ratios between total and CD44-expressing CD8+ and CD4+ T cells, were not altered in tumors from NOS2KO animals (Supplementary Fig. S4D). The use of anti-CTLA and anti–PD-1 checkpoint blockade therapy 10 days after tumor injections abrogated tumor growth differences in WT and NOS2KO animals (Supplementary Fig. S4E and S4F).
These results demonstrate the critical role played by endogenous synthesis of NO in T cells, particularly during antitumor immune responses.
Endogenously synthesized NO facilitates CD8+ T-cell homing to tumors
We next employed a mouse model of adoptive cell therapy (ACT) to assess the function of tumor specific NOS2KO OT-I cells compared with WT OT-I cells (Fig. 5A). WT mice were subcutaneously inoculated with B16-F10-OVA cells, lymphodepleted with CPA, and intraperitoneally injected with 4-day activated WT or NOS2KO OT-I cells. A group of animals did not receive T cells (No ACT). Tumor growth was measured for 60 days following ACT and showed that animals injected with NOS2KO OT-I cells had faster tumor growth and reduced survival compared with WT controls (Fig. 5B). OT-I–receiving animals exhibited delayed tumor growth relative to No ACT controls.
We also further assessed the ability of NOS2KO CD8+ T cells to infiltrate tumors relative to WT CD8+ T cells (Fig. 5C). OVA-expressing B16-F10 melanoma cells were subcutaneously injected into WT mice, followed by lymphodepletion with CPA, and adoptive transfer of a cell suspension containing naïve WT and NOS2KO OT-I cells at a 1:1 ratio. OT-I tissue infiltration was analyzed by flow cytometry of single-cell suspensions prepared from spleen, blood, liver, and tumors 5 days after cotransfer of T cells to tumor-bearing animals. Endogenous and adoptive populations were distinguished by the allelic variants of CD45 (Supplementary Fig. S5A). A higher proportion of NOS2KO OT-I cells was found in nonmalignant tissues (spleen, blood, and liver) relative to WT OT-I cells, whereas significantly fewer NOS2KO OT-I cells were detected in tumors (Fig. 5D; Supplementary Fig. S5B). Deletion of Nos2 resulted in a reduced proportion of CD8+T cells expressing the effector molecule granzyme B (Fig. 5E). In all tissues analyzed and relative to WT controls, NOS2KO cells expressed higher levels of CD8 and CD3, which are typically downregulated after T-cell activation (Supplementary Fig. S5C). Lack of Nos2 also decreased the proportion of tumor-specific (OT-I) CD44+ cells circulating in the blood (Fig. 5F).
To better understand response of T cells to endogenous NO production, we also assessed the antitumor function of OT-I cells overexpressing Nos2 in an ACT model (Supplementary Fig. S5D). WT mice were subcutaneously inoculated with B16-F10-OVA cells, lymphodepleted with CPA, and intraperitoneally injected with 4-day activated OT-I cells transduced with retroviral vectors and purified with Thy-1.1 beads by MACS. Experimental groups included animals not receiving T cells (No ACT), animals receiving OT-I cells transduced with VC, and animals receiving OT-I cells transduced with NOS2OE vectors (NOS2OE). Blood sampling on day 14 following ACT showed that the frequency of NOS2OE cells in circulation was extremely low compared with control cells (Supplementary Fig. S5E). Tumor growth was measured for 50 days following ACT and showed that both VC and OT-I receiving animals delayed tumor growth relative to No ACT and NOS2OE groups (Supplementary Fig. S5F). Tumor growth in animals receiving NOS2OE OT-I cells resembled that of No ACT animals.
These data showed that, similar to exposure to high NO from an exogenous source, endogenous high production of NO negatively impacted T-cell expansion and function. However, ablating endogenous production of NO reduced the ability of T cells to fully differentiate, infiltrate, and efficiently eradicate tumors.
Decreased NO production impacts in vivo CD8+ T-cell differentiation and recall response
We next investigated whether in vivo T-cell activation and recall responses, key for antitumor T-cell function, were affected by the loss of Nos2 (Fig. 6A). Naïve NOS2KO and WT OT-I CD8+ T cells were cotransferred into WT recipient mice in a 1:1 ratio (Supplementary Fig. S6A and S6B). One day later, T-cell stimulation was performed using SIINFEKL-pulsed BMDMs. The ratio of NOS2KO to WT CD8+ T-cell counts in peripheral blood on days 7 and 10 was used to track the activation and expansion of both populations. On day 30, a recall response was triggered by injecting mice with a second dose of SIINFEKL-pulsed BMDMs, and 7 days later, the ratio of NOS2KO to WT cell counts within different organs was used to determine the relative fitness of NOS2KO cells during reactivation. Endogenous and adoptive populations were distinguished by the allelic variants of CD45 (Supplementary Fig. S6A).
On day 7 after transfer, NOS2KO OT-I cells showed a reduced proportion of terminally differentiated cells (CD44+CD62L−) and higher expression of CD8 and CD127, which is commonly associated with less differentiated T cells (Fig. 6C; Supplementary Fig. S6C). Ten days after adoptive transfer, fewer NOS2KO cells were found in peripheral blood (Fig. 6B; Supplementary Fig. S6C). In the absence of an antigenic recall response, the ratio between NOS2KO and WT OT-I cells was close to 1 in the spleen and lymph nodes, whereas fewer NOS2KO T cells were found in the liver relative to WT cells (Fig. 6D). Recall with antigen-pulsed BMDMs led to expansion of total and CD44+ OT-I cells in the spleen, lymph node, and liver relative to animals receiving PBS injections instead of BMDMs (Fig. 6E; Supplementary Fig. S6D). However, NOS2KO T cells were reduced in all tissues relative to cotransferred WT cells (Fig. 6E; Supplementary Fig. S6D). The spleen was the only tissue where recall responses significantly reduced the ratio of NOS2KO:WT compared with PBS controls (Fig. 6D), whereas only in the liver were the levels of CD44+ OT-I cells reduced in NOS2KO cells relative to WT OT-I in the PBS controls (Fig. 6E). In all tissues analyzed after recall, NOS2KO OT-I infiltrates had lower CD44 expression compared with WT controls (Fig. 6F).
These data demonstrate that the endogenous synthesis of NO is an essential aspect of CD8+ T-cell differentiation, expansion, and recall responses in vivo.
The generation of NO in the immune system has generally focused on T cells as passive players, where myeloid or tumor cell synthesis has been shown to be a significant suppressor of T-cell activity. Although these data are compelling and borne out by our own (8) and others studies (34) of hypoxia-induced myelosuppression of CD8+ T-cell proliferation, it is also clear that NO can be used by different cell types to carry out different processes. In that regard, the data we present here provides evidence that NO is not solely inhibitory, as it can also significantly augment adaptive antitumor immune responses.
In this study, we generated a T cell–specific NOS2KO mouse model. CD8+ T cells from this model showed significantly decreased synthesis of NO, which was particularly evident both after activation and in reduced oxygen settings. Previous studies of T-cell function and NOS2 have used whole animal deletion models, which are to some extent compromised by the effects of deletion of the gene in myeloid lineages (5, 22). Thus, the work presented here allowed an isolated and focused approach to the importance of NOS2 expression specific to T cells.
In unchallenged animals, the immune populations in the thymus, spleen, blood, and lymph nodes of Nos2fl/fldlckCre (NOS2KO) animals resembled those of Nos2fl/fl (WT) controls. This was likely in part due to the use of a Cre recombinase promoter, which is only active in mature T cells (dLck), as the lack of NOS2 in double-positive thymocytes has been found to impair their selection (35). However, tumor growth in two different models, B16 melanoma and MC38 colon carcinoma, showed significantly increased progression and resulted in decreased survival rates in animals lacking Nos2 in the T-cell compartment. We further assessed the functional role of T cell–derived NO using tumor specific CD8+ OT-I cells with altered NO synthesis. Augmenting intracellular NO with Nos2 overexpression was detrimental to the function of adoptively transferred CD8+ T cells. Similar to the effect of exogenous NO, derived from myeloid cells (8) or NOC-18, the rise in intracellular NO in Nos2-overexpressing cells suppressed T-cell expansion, which could explain the impaired tumor growth control in vivo. Decreasing intracellular NO synthesis with Nos2 deletion also impaired antitumor function of cells following adoptive cell transfer. We found that this might be explained by altered CD8+ T-cell differentiation, recall responses, and tissue infiltration.
After TCR stimulation, either in vitro or in vivo, NOS2KO CD8+ T cells had decreased expression of several differentiation markers (CD44, CD25, and ICOS) and higher levels of CD8 and CD127, both known to be downregulated with T-cell activation (36, 37). In vitro, these results were only observed after culture at 1% O2, keeping with the higher expression of NOS2 we and others have observed in low oxygen (33, 38, 39). In these same conditions, NOS2KO CD8+ T also showed reduced expression of the effector molecules granzyme B and IFNγ, as well as changes in expression of T-bet, a key transcription factor in effector T-cell responses, thus arguing endogenous expression of NO plays a role in effector T-cell function. When compared with their WT counterparts, NOS2KO CD8+ T cells retained the same proliferative capacity, cytotoxic function, and OCR, contrary to what has been shown, for example, in macrophages lacking Nos2 (40–42). As previously shown (12, 20), activation of NOS2KO CD4+ T cells increases the proportion of Th17−expressing cells following Th17 polarization.
In mouse models where NOS2 mutant and WT T cells were cotransferred into recipient mice, the lack of NOS2 in T cells reduced their expansion in the blood and significantly reduced the recall response. The lack of NOS2 in T cells also significantly reduced transendothelial migration in vitro and tumor infiltration in vivo. Infiltration of both CD4+ and CD8+ T cells was decreased in B16-bearing Nos2fl/fldLckCre animals, indicating that endogenous NO can modulate both CD8+ and CD4+ T cells. When compared with WT cells, NOS2KO CD8+T lymphocytes showed a reduced expression of the effector molecule granzyme B and increased expression of CD8 and CD3, which are downregulated following T-cell activation. This indicates that endogenous production of NO contributes to both to T-cell differentiation and homing of cytotoxic T cells to tumors. In fact, NOS2 expression by T cells has been shown to influence CD3+ cell tissue infiltration and vascular dysfunction of human allografts, which was partially reverted with 1400W treatment (43).
Increasing NO production in CD8+ T cells by overexpressing Nos2 (NOS2OE) significantly reduced cell proliferation; this was abrogated by treatment with a NOS2 chemical inhibitor, 1400W. Although CD44 was downregulated in NOS2KO T cells, it was increased in NOS2OE CD8+ T cells, and this effect could also be reversed by 1400W. Within the population of cells transduced with the NOS2OE vector, only the NOS2high cells overexpressed CD44. This indicates that endogenous production of NO correlates with CD44 expression. As hyaluronic acid, the CD44 ligand, was shown to increase NO production in chondrocytes and endothelial cells, this dose-dependency of CD44 expression on endogenous NO could be explained by a positive feedback loop controlling CD44 ligand binding and cell trafficking (44, 45). Given the role of CD44 in transendothelial migration of lymphocytes (46), the reduced expression of CD44 in NOS2KO CD8+ T cells might in part explain the lower migration capacity of NOS2 mutant cells through an endothelial cell barrier, and their altered tissue infiltration pattern in vivo when compared with WT cells.
As seen in murine T cells, human CD8+ T cells produced NO, primarily through the expression of NOS3. Pharmacologic inhibition of NO production with L-NAME decreased expression of the trafficking molecules CD25 and T-bet in human T cells in a manner similar to that observed in murine T cells following Nos2 deletion. These cross-species similarities highlight the importance of endogenous NO signaling in T cells.
Our data provide evidence that NO is endogenously synthesized by T cells at low levels and has clear immunomodulatory effects on their differentiation, recall response, and tissue infiltration, with the potential to affect antitumor immunity. The data indicate that finely tuned modulation of the NO synthesis pathway might allow for improved T-cell function in immunotherapeutic settings.
No disclosures were reported.
P.P. Cunha: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. D. Bargiela: Investigation, writing–review and editing. E. Minogue: Investigation, writing–review and editing. L.C.M. Krause: Investigation. L. Barbieri: Investigation. C. Brombach: Investigation. M. Gojkovic: Investigation. E. Marklund: Investigation. S. Pietsch: Investigation. I. Foskolou: Supervision, investigation, writing–review and editing. C.M. Branco: Conceptualization, formal analysis, supervision, funding acquisition, investigation, writing–original draft, project administration, writing–review and editing. P. Veliça: Conceptualization, formal analysis, supervision, investigation, writing–original draft, project administration, writing–review and editing. R.S. Johnson: Conceptualization, resources, formal analysis, supervision, funding acquisition, investigation, writing–original draft, project administration, writing–review and editing.
The work was funded by the Knut and Alice Wallenberg Scholar Award, the Swedish Medical Research Council (Vetenskapsrådet 2019–01485), the Swedish Cancer Fund (Cancerfonden, CAN2018/808), the Swedish Children's Cancer Fund (Barncancerfonden PR2020–007), the Portuguese Foundation for Science and Technology scholarship to Pedro P. Cunha (SFRH/BD/115612/2016), a Newnham College research fund to Cristina Branco, and the Principal Research Fellowship (214283/Z/18/Z) to Randall S. Johnson from the Wellcome Trust.
The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.
Note: Supplementary data for this article are available at Cancer Immunology Research Online (http://cancerimmunolres.aacrjournals.org/).