Jeko-1, Mino, MM.1S, and HEK293T cells were bought from the American Type Culture Collection (ATCC), and DAOY cells were a gift from Dr. Alex Huang (Case Western Reserve University, OH, USA). Jeko-1, Mino, MM.1S, and DAOY cells were cultured in RPMI-1640 (Sigma-Aldrich) supplemented with 10% fetal bovine serum (FBS) (Sigma-Aldrich) and 1% Pen/Strep (Hyclone). HEK293T cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) (Sigma-Aldrich) with the same 10% FBS and 1% Pen/Strep supplementation. NK92 cells were a kind gift from Dr. Daniel Popkin (Innova Dermatology, TN, USA). They were cultured in Alpha-MEM media (Sigma-Aldrich) and supplemented with 12.5% FBS, 12.5% horse serum (Sigma-Aldrich), 0.02 mM folic acid (Acros), 0.1 mM 2-mercaptoethanol (Gibco), 1.5 g/L sodium bicarbonate (Fisher Scientific), 2 mM L-glutamine (Gibco), and 0.2 mM inositol (Acros). IL-2 (200 IU) was added fresh ever 2–3 days with media changes for NK92 cells. All cells were maintained at 37°C with 5% CO₂ in the cell incubator. Glucose concentrations of media used for cell culture: 2 g/L for primary NK cells, as they were in RPMI-1640 media; 1g/L for NK92 cells, as they were cultured in Alpha-MEM; and 4.5 g/L for HEK293T cells, as we use DMEM-High Glucose. All media are from Sigma-Aldrich as mentioned above.

Healthy donors were recruited via the Hematopoietic Stem Cell Core Facility at Case Western Reserve University, and 40 ml of blood in EDTA was collected per donor. Blood was isolated via Ficoll (GE Healthcare) gradient centrifugation to isolate peripheral blood mononuclear cells (PBMCs). Here, 1 × 10⁷ PBMCs were plated with supplemented RPMI media, IL-2 at 100 U/ml, and K562-irradiated feeder cells. IL-2 was replaced every 3 days, and feeder cells were replaced every 7 days. Experiments were run with cells ranging from Day 10 to Day 21. For certain experiments, IL2 dosing was increased up to 300IU as indicated in respective figure legends.

Phosphate-buffered saline (PBS), untreated NK cells, or OSMI-1 (25 μM)-treated primary NK cells were utilized. Cells were incubated with or without OSMI-1 for 36 h. Primary NK cells were utilized from one of 2 donors on given injection days. Cells were only injected if they were between Day 9 and Day 17 post-isolation from the donor. Cells were washed 2x with PBS at 900 revolutions per minute (rpm) to remove OSMI-1 from media and injected as a cell suspension in PBS. Dilution required for 3 × 10⁶ cells was determined via Countess (Invitrogen) hemocytometer, and 100 µl of suspension was injected per mouse.

Mice were bred at the Case Western Reserve University in accordance with the guidelines of the Institutional Animal Care Use Committee. NOD SCID IL-2r Gamma (NSG) mice were purchased from Case Western Reserve University Athymic Animal Core. Experiments were conducted using age- and gender-matched mice in accordance with approved institutional protocols. Mice were kept in their permanent housing at the imaging core facility for a minimum of 1 week before any experiments were performed.

Mice were anesthetized via isoflurane chamber with subsequent transfer to nose cone. Here, 1 × 10⁷ Jeko-luciferase-tagged cells were injected subcutaneously at the dorsal right flank. On days of imaging, mice were injected intraperitoneally (IP) with 150 mg/kg D-luciferin (PerkinElmer). After 8 days, initial tumors were measured via spectrum IVIS imaging system (PerkinElmer). Subsequent measurements occurred every 3-4 days. As indicated, primary NK cells isolated from one of two donors were injected at 3 × 10⁶ NK cells with 50 ng/mouse of human rhIL-15 (BioLegend). Luminescent images were utilized from 10 min post-injection of D-luciferin, and images were analyzed by Living Image software (PerkinElmer).

pLKO and pLKO.1 shRNA plasmids targeting human OGA (TCRN0000133855; Mission Sigma) or human OGT (shOGT1-TRCN0000286200, shOGT2-TRCN0000286199, and shOGT3-TRCN0000293652; Mission Sigma) or green fluorescent protein (GFP) (Plasmid SHC005; Sigma) were cotransfected with lentivirus-packaging plasmids psPAX2 (Plasmid 12260; Addgene) and pMD2.G (Plasmid 12259; Addgene) at a ratio of 4:1:3 using X-treme GENE HP DNA transfection reagent (Sigma). HEK293T cells used X-treme Gene Transfection Reagent per protocol. Viral supernatants were collected at 48 and 72 h posttransfection. For viral transduction, 1 × 10⁶ cells were resuspended in 1 ml virus-containing medium with 10 µg/ml of polybrene into 1 well of a 12-well plate and centrifuged at 3,480 rpm for 90 min in a centrifuge that was prewarmed at 32°C. Here, 1 ml of fresh supplemented Alpha-MEM media were added to the cells per well, and cells are resuspended into a tissue culture flask of appropriate volume. The following morning, the cell culture medium is changed. Forty-eight hours later, selection with 2 μg/ml puromycin was started for 6 days. After 6 days, selection pressure was removed and cells were cultured per cell line and culture section. For HEK293T cell experiment determining shRNA plasmid efficacy: HEK293Ts were transfected at above ratio, the medium was changed the following morning, and O-GlcNAc levels were determined by flow cytometry 36 h posttransfection.

Cancer cells were stained with efluor 670 according to protocol. NK92 or primary NK cells were incubated for 24 h without treatment or with OSMI-1/TG. The cells were then spun down at 900rpm for 5 min, and 160 μl of supernatant from treated groups was added to wells of a 96-well Falcon U-bottom plate. Here, 40,000 efluor 670-stained cancer cells were added to each well. Moreover, 120,000 NK cells were added to each well. Cells were incubated overnight, and in the following morning, cells were spun down, washed, and stained with propidium iodide (PI). A total of 10,000 allophycocyanin (APC)-positive events were collected, and the experiment was performed in triplicate. Supernatant from overnight incubation of cancer cells and NK cells was collected after the first spin prior to washing and PI staining. This supernatant was utilized for LEGENDPlex NK/T Cell Panel Analysis (BioLegend) per protocol. LEGENDPlex values from flow cytometry were back calculated to pg/ml utilizing standard curve interpolation on Prism (see Statistical Methods section).

Cells were washed with PBS and stained with PI, which was subsequently analyzed via the PE channel on flow cytometry. Alternatively, cells were stained with trypan blue (Sigma) and analyzed via Countess (Invitrogen) hemocytometer. Counts were performed in triplicate.

CD69 data collection and analysis were performed using the Accuri C6 flow cytometer and software (BD Biosciences). Cytoflex (Beckman-Coulter) was utilized for data collection of receptor, cytosolic O-GlcNAc expression, cytotoxicity, viability, and cytokine experiments. Analysis was performed on FlowJo software. The following antibodies (BioLegend) were used with listed conjugations and clones in parentheses: CD3 (FITC, HIT3a), CD56 (APC and FITC, 5.1H11), KIR2DL1 (PerCP/Cy5.5, HP-MA4), KLRG1 (PE, 14C2A07), NKG2A (PE, S19004C), NKp30 (PE, p30-15), NKp44 (PerCP/Cy5.5, p44-8), and NKG2D (PerCP/Cy5.5, 1D11). IgG1 Kappa Isotype (Alexafluor647, P3.6.2.8.1), and O-GlcNAc (Alexafluor 647, RL2) flow antibodies were purchased from Thermofisher Scientific. shOGA and shGFP cells were not examined until at least 7 days after puromycin selection. Intracellular O-GlcNAc levels were examined by flow cytometry every other week and confirmed to still be significantly elevated in shOGA before any other downstream analysis was performed. Intracellular staining was performed using Foxp3 intracellular staining kit (00-5523-00, Thermo Fisher). For the experiment, free GlcNAc (A3256-25G, Sigma-Aldrich) competed with the anti-O-GlcNAc antibody mentioned above for binding. NK cells were isolated and incubated with the antibody for 45 mins with or without 50mM, 250mM, and 500mM GlcNAc, and O-GlcNAcylation detection was assessed via flow cytometry.

Protein was isolated with RIPA buffer (Sigma) with protease and phosphatase single-use inhibitor (78442, Thermo Fisher). Western blots were performed using Bio-Rad mini protean system. Depending on the experiment, 20–35 µg of protein was loaded per well and diluted with 4× Laemmli buffer (Bio-Rad). Beta-actin was used as a loading control (SC-47778, Santa Cruz). O-GlcNAc antibody (ab2739, Abcam) was utilized. Chemiluminescence was performed with systems from Pierce according to protocol and imaged on Bio-Rad Chemidoc XRS+ and analyzed on Quantity One (Bio-Rad). Figure 1 Western blots were performed in a dark room with X-ray film and SRX-101A developer (Konica Minolta). Antibodies used in Figure 1 and developed in the dark room are as follows: GADPH (sc47724), GFAT1 (sc377479), GNA1 (sc374519), PGM3 (sc390239), OGT (sc32921) from Santa Cruz, and UAP1 (ab95949) from Abcam. Antibodies used in Supplementary Figure S2 are as follows: NF-κB (8242T), NFAT1 (5861T), p-AKT at T308 (13038T), pan AKT (4691), ERK/p42/p44 (4695T), p-ERK T202/Y204 (9101S), PLCgamma2 (3872T), STAT5 (25656), and p-STAT5 Y694 (9351S)—all from Cell Signaling Technology normalized to beta-actin from Santa Cruz mentioned above. Western blots were quantified using ImageJ software. Area under the curve was determined and normalized to respective loading control’s [beta actin or Glyceraldehyde-3-Phosphate Dehydrogenase (GAPDH)] area under the curve.

Statistical analysis was conducted via GraphPad Prism software version 9. Statistical tests described in figure legends were used for respective experiments. Asterisks were used to display varying levels of significance from p < 0.05 to p < 0.0001 as described in figure legends. A summary of statistical methods used is as follows: Student’s unpaired T-test and one-way ANOVA with Dunnett’s multiple comparisons to the NK group mean or HEK293T cells that were not transfected. Standard curve creation for LEGENDPlex Analysis was performed utilizing sigmoidal 4PL, X is concentration as interpolation setting. In vitro experiments were performed 2–4 times with at least 3 technical replicates per experiment. In vivo experiments were performed with the number of mice indicated by (n) per group.

O-GlcNAcylation is the end product of the HBP, where glucose is converted to UDP-GlcNAc and OGT transfers GlcNAc from UDP-GlcNAc to substrate proteins and OGA removes it ( Figure 1A ). We examined the expression levels of enzymes involved in the HBP pathway before and after stimulation with IL-2 in primary NK cells. As shown in Figure 1B and Supplementary Figure S1A , OGT, GlcNAc phosphomutase (PGM-3), glutamine fructose-6-phosphate amidotransferase-1 (GFAT-1), UDP-N-acetylglucosamine pyrophosphorylase (UAP-1), and glucosamine-6-phosphate N-acetyltransferase (GNA-1) levels were increased in NK cells activated by IL-2. We also confirmed that the viability of NK cells with and without the cytokine is comparable ( Figure 1C ). To confirm whether IL-2 treatment increases O-GlcNAcylation, we performed Western blotting and found an increased level of global O-GlcNAcylation after 24 h of stimulation of freshly isolated NK cells with IL-2. Similarly, an increase in O-GlcNAcylation was noticed after 48 h of IL-2 or IL-15 treatment of cultured expanded NK cells ( Figure 1D and Supplementary Figure S1B ). Activation of NK cells after cytokine stimulation was confirmed by an increase in CD69 expression, a marker of NK cell activation ( Figure 1E ). This shows that an increase in O-GlcNAcylation is associated with cytokine-induced NK cell activation.

OSMI-1 and TG are well-established inhibitors of OGT and OGA, respectively. We measured O-GlcNAc protein levels by Western blotting and found marked reduction of O-GlcNAc levels with OSMI-1 (25 μM) and marked increase of O-GlcNAc levels with TG (50 μM), both after 24 h of incubation ( Figure 1F , Supplementary Figure S1C ). Next, we examined the decrease or increase in intracellular O-GlcNAcylation by flow cytometry after incubation of cells with OSMI-1 and TG, respectively, using an anti-O-GlcNAcylated flow cytometry antibody (RL-2) ( Figures 1G, H ). To ensure that this antibody specifically binds only to intracellular O-GlcNAcylated proteins, we tested its binding to surface proteins on NK cells and found no detection of O-GlcNAc signal on the cell surface of NK cells ( Figure 1I ). To further validate this antibody, a competition assay was performed by adding free GlcNAc at increasing concentrations, 50, 250, and 500 mM, for 45 min during the staining with RL2 antibody, and O-GlcNAc levels were detected by flow cytometry. We found a decreased detection of O-GlcNAc even at the lowest tested concentration, 50 mM GlcNAc, which further went down in the presence of 250 and 500 mM free GlcNAc ( Figure 1J ), confirming the specificity of RL-2 antibody in detecting O-GlcNAc by flow cytometry.

Cell surface receptors expressed on NK cells act as a sensory system that initiates downstream signaling after binding to its respective ligands, which determines the cellular response of NK cells. To determine whether alteration of O-GlcNAcylation in NK cells results in changes in receptor expression, we treated primary NK cells with OSMI-1 or TG for 24 h and measured surface receptor expressions via flow cytometry. We show that surface expressions of activating receptors NKp44 and NKG2D and inhibitory receptor NKG2A are significantly downregulated ( Figures 2A, B ), inhibitory receptor KIR2DL1 expression is upregulated, while KLRG1 expression remains unaltered after OSMI-1 treatment ( Figures 2C, D ). HLA-E, ligand of NKG2A receptor, expression on tumor cells was detected, and 2 out of 3 tumor cells showed very low expression ( Supplementary Figure S2A ). Analysis of the cell signaling pathways [phospholipase C-gamma-2 (PLC-γ2), STAT5, extracellular signal-regulated kinase 1/2 (ERK1/2), AKT] and transcription factors [nuclear factor-κb (NF-κb), Nuclear factor of activated T-cells (NFAT)] involved in NK cell function showed some minor alterations after OSMI-1 treatment ( Supplementary Figures S2B–E ) that needs further confirmation using NK cells from multiple donors. Cytokines secreted by NK cells also play a crucial role in antitumor activity. To determine the cytokine secretion levels, primary NK cells were treated as above and culture supernatants were analyzed via LEGENDplex (BioLegend). IL-17 levels did not show any changes ( Figure 2E ), while decreased levels of TNF-α, IFN-γ, soluble Fas ligand, perforin, granzyme A, granzyme B, and granulysin were observed after OSMI-1 treatment ( Figure 2F ). Interestingly, TG treatment did not cause significant alterations in receptor or cytokine secretion after 48 h ( Figures 2A–F ).

Given the changes in surface receptor profile and cytokine secretion, we investigated whether NK cell’s antitumor cytotoxicity was affected following OSMI-1 treatment. Primary NK cells were incubated with OSMI-1 or TG for 24 h and then coincubated at a 3:1 ratio with efluor-670-stained mantle cell lymphoma (MCL) cell lines Jeko and Mino or multiple myeloma (MM) cell line MM.1S or medulloblastoma cell line DAOY. After 12 h of coincubation for hematologic malignancies and 4 h of coincubation for the solid tumor DAOY, we analyzed the percent tumor cell death. An example of the cytotoxicity gating on APC (efluor-670)-stained cancer cells and their PI uptake is shown in Figure 3A . This experiment showed substantial and statistically significant decreases in cytotoxicity (30%–65%) by NK cells after OSMI-1 treatment compared to control NK cells. No significant alterations in cytotoxicity were seen in NK cells post TG treatment ( Figure 3B ). We checked the viability of NK cells themselves after 36 h with OSMI-1 or TG treatment and found no major differences in cell viability after OSMI-1/TG treatments ( Figure 3C ). Finally, we also examined whether OSMI-1 and TG have any direct effects on cancer cells. No major cell death was noticed after OSMI-1 or TG treatment for 12 h ( Figure 3D ).

It is challenging to generate stable cells after viral transductions of primary NK cells (26). Hence, we chose the NK cell line NK.92 to do genetic modifications. First, we wanted to confirm if OSMI-1 and TG treatments had similar effects in NK.92 cells as seen in primary NK cells. We examined 3 receptors, NKG2D, NKG2A, and KIR2DL1 and found that the expression of NKG2D and NKG2A decreased ( Figure 4A ) while KIR2DL1 increased ( Figure 4B ) after OSMI-1 treatment, similar to primary NK cells. We found no significant changes in expression of these receptors after TG treatment ( Figures 4A, B ). In addition to this, a significant reduction in NK.92 cell cytotoxicity after OSMI-1 treatment was observed after coincubation with two different cancer cells Jeko-1 and MM1.S ( Figure 4C ).

We validated one shOGA and three shOGT plasmids by transducing HEK293T cells. shOGA-transduced cells showed a significant increase and shOGT-transduced cells showed a significant decrease in O-GlcNAcylation as measured by flow cytometry ( Figure 4D ). We then created a stable NK.92 cell line expressing shOGA or shGFP as a control. We found a substantial increase in the O-GlcNAcylation levels in shOGA cells compared to shGFP control cells ( Figures 4E, F ; Supplementary Figures S3A–C ). We used the shOGA cell line to validate the data obtained after TG treatment in NK.92 cells. We determined whether shOGA transduction in NK.92 cells resulted in alterations in its cytotoxic function. No alterations in cytotoxicity were seen between the shGFP and shOGA stable cell lines when cocultured with Jeko or MM.1S cancer cells ( Figure 4G ). Next, we checked the surface receptor analysis by flow cytometry. We did not see any major changes in NKG2A, KIR2DL1, and KLRG1 receptor expression ( Figure 4H ). These shOGA-transduced cells behaved similarly to the TG-treated NK cells and confirm that the increase in O-GlcNAcylation in NK cells results in no major changes in cytotoxic activity by using two approaches.

Our attempts to create shOGT stable knockdown were unsuccessful, despite successful generation of shGFP- and shOGA-transduced NK.92 cell lines. shOGT-transduced NK.92 cells died after puromycin selection, while shGFP and shOGA cells were viable as evident from PI staining shown in Figure 4I and Supplementary Figure S3C . This implies that O-GlcNAcylation is required for NK92 cell viability.

We used a mantle cell lymphoma, Jeko-1 cell, xenograft model to study in vivo cytotoxicity of NK cells pretreated with OSMI-1 compared to control NK cells. NSG mice were injected with Jeko cells expressing luciferase subcutaneously and NK cells (control NK or NK cells pretreated with OSMI-1) were injected intratumorally on days 13, 16, and 19, and mice were sacrificed on day 21. A schematic of the experiment is shown in Figure 5A . Primary NK cells (CD3^- CD56⁺) ( Figure 5B ) were tested to ensure good viability 15 min before injection and that viabilities were not significantly different in control and OSMI-1-treated groups after 36 h of treatment ( Figure 5C ). Luciferase image quantification at Day 21 is shown as radiance ( Figure 5D ). Tumor growth evaluated by radiance is charted through Day 21, and imaging was done twice weekly. This imaging shows that NK cell-treated mice had overall reduced tumor burden while tumors progressed in mice injected with OSMI-1-treated NK cells similar to PBS-injected mice ( Figure 5E ). Luciferase mouse imaging at Day 21 is shown ( Figure 5F ). Overall, the in vivo data validated the in vitro data that OSMI-1-mediated suppression of O-GlcNAcylation inhibits NK cell cytotoxicity.