Lysosomal NKG7 restrains mTORC1 activity to promote CD8⁺ T cell durability and tumor control

Abstract

During infection and cancer, mTORC1-mediated metabolic regulation impacts CD8⁺ T cell effector expansion and memory development. However, the mechanisms by which CD8⁺ T cells regulate mTORC1 to support their unique metabolic requirements remain unknown. Here we show that NKG7, a lysosomal protein whose expression is restricted to cytotoxic lymphocytes, negatively regulates mTORC1 recruitment and activation by inhibiting assembly and function of the lysosomal proton pump, vacuolar ATPase (v-ATPase). Human and mouse CD8⁺ T cells lacking NKG7 show more acidic lysosomes and increased activation of mTORC1 signaling, which could be reversed by inhibition of v-ATPase activity. In mice responding to LCMV infection, NKG7-deleted effector CD8⁺ T cells are less durable and generate fewer memory precursors, whereas induced expression of NKG7 in CD8⁺ T cells results in increased presence of intra-tumoral T cells. Overall, our work identifies NKG7 as a CD8⁺ T cell-specific regulator of mTORC1 activity, required for optimal immune responses.

Although effector activity and memory formation in CD8 + T cells are known to depend on mTORC1-mediated metabolic regulation, the molecular mechanisms involved are lesser known. Here authors show that NKG7, a lysosomal protein specifically expressed in CD8 + T cells, inhibits mTORC1 function via the lysosomal proton pump, vacuolar ATPase, promoting antitumor activity and expansion of memory T cell precursors.

Introduction

The lysosome is an intracellular organelle that sits at the crossroads of both CD8⁺ T cell cytotoxicity and metabolism^1–3. While well-defined as a central hub for macromolecular degradation and recycling, the lysosome has a much broader impact on cellular homeostasis through its regulation of membrane repair, cell motility, adhesion, autophagic flux and metabolic signaling^4–7. Moreover, the lysosome is critical to the generation of lytic granules, which contain the cytolytic molecules (perforin and granzymes) needed for killing by CD8⁺ T cells^8,9. However, the role of the lysosome in T cell metabolism has not been clearly defined as it pertains to balancing T cell cytolytic potential and durability.

The mechanistic target of rapamycin (mTOR) complex 1 (mTORC1), a highly conserved serine/threonine kinase complex, is a master regulator of cellular metabolism in eukaryotes. Upon recruitment and activation at the lysosome, mTORC1 promotes anabolic metabolism by integrating signals from extracellular stimuli, intracellular metabolites, and stress. Consequently, mTORC1 plays a critical role in determining cellular growth, survival, and fate in all types of cells, and mTORC1 dysregulation is associated with assorted human diseases^10,11. Although the mechanisms by which extra- and intracellular signals converge on mTORC1 activation have been elucidated, whether and how this central metabolic machinery is further fine-tuned and optimized in a cell type-specific manner has remained elusive. This is especially true in the case of CD8⁺ T cells, which constantly encounter dynamically changing microenvironments (circulation vs tissue/tumor environment)^12,13.

Cytotoxic CD8⁺ T cells are essential for the elimination of viral-infected and tumor cells. Thus, the success of most cancer immunotherapies relies on efficient tumor cell killing mediated by CD8⁺ T cells^14,15. Dynamic reprograming of the metabolic state within CD8⁺ T cells is necessary for cellular activation, effector function and memory generation, and metabolic profiles associated with each CD8⁺ T cell status have been described^12,16. However, the intracellular pathways by which CD8⁺ T cells transition into different metabolic modes according to their intrinsic cues or integrate external cues differentially to meet their intrinsic metabolic requirements is unclear. Complete knowledge of these regulatory steps is still needed to define how CD8⁺ T cells can maintain their cytolytic potential, survival and memory differentiation when exposed to harsh microenvironments harboring chronic viral infections and tumors, where nutrients might be limiting^17,18.

In search of genes whose expression is linked to a durable CD8⁺ T cell response to immune checkpoint inhibitor (ICI) therapy in human cancer patients, we previously found that CD8⁺ T cells from ICI responders had higher levels of NKG7 (natural killer cell group 7; also known as granule membrane protein-17 [GMP-17]) expression when compared to ICI non-responders¹⁹. NKG7 is a 4-transmembrane spanning protein discovered 30 years ago that localizes to the lysosome and is mostly expressed by cytotoxic immune cells, primarily CD8⁺ T cells and NK cells^20,21. The importance of NKG7 in regulating lytic granule secretion and anti-tumor immunity has been shown by our group and others^22,23. However, whether NKG7 has other functions in addition to the regulation of cellular cytotoxicity are unknown.

In this study, we show that NKG7 negatively regulates mTORC1-mediated metabolic processes in CD8⁺ T cells, by impairing mTORC1 recruitment to lysosomes. Mechanistically, we demonstrate that NKG7 interacts with and inhibits the lysosome assembly of the vacuolar ATPase (v-ATPase), which impacts recruitment and activation of mTORC1. Physiologically, NKG7-mediated mTORC1 negative regulation is required for the generation of long-lived CD8⁺ T cell responses to acute or chronic viral infection and increased tumor infiltration and survival. NKG7, therefore, is a gatekeeper of mTORC1 activity in effector CD8⁺ T cells, providing a necessary regulatory step for the establishment of CD8⁺ T cell durability and memory generation in response to infections and solid tumors.

Results

The loss of NKG7 increases the formation of late endosome – lysosome hybrid organelles

To examine whether NKG7 affects lysosome homeostasis, we first examined lysosomal morphology and distribution upon NKG7 depletion. We observed that NKG7 knockout (KO) in both primary human CD8⁺ T cells and NK cells resulted in more aggregated/enlarged lysosomes (LAMP1⁺) with a concomitant decrease in lysosome numbers (Fig. 1a–c and Supplementary Fig. 1a). Our result is consistent with a previous observation of lysosomes in Nkg7-KO mouse CD8⁺ T cells²⁴. Conversely, when NKG7 is over-expressed (OE) in Jurkat CD4⁺ T cells or HEK293T (293 T) cells, which do not express NKG7 endogenously, we observed the presence of smaller lysosomes compared to the control group with a concomitant increase in lysosome number and more dispersed distribution (Fig. 1d–f and Supplementary Fig. 1b). We next examined whether NKG7-mediated lysosomal morphology and/or distribution is through regulation of lysosomal trafficking along the microtubule (MT) network. Lysosomal distribution is determined by coordination between dynein-dynactin-mediated retrograde and kinesin (kinesin-1 and -3)-mediated anterograde trafficking along the MT network²⁵. Interestingly, we observed that disruption of the MT network in CD8⁺ T cells via nocodazole treatment resulted in aggregated/enlarged lysosomes phenocopying NKG7-KO cells (Supplementary Fig. 1c, d). However, distribution of both anterograde and retrograde MT motors (kinesin family member 5B [KIF5B] and dynein intermediate chain [DIC], respectively) involved in lysosomal trafficking were not affected upon depletion of NKG7 (Supplementary Fig. 1e–h). Therefore, we considered an alternative possibility that NKG7 could regulate fusion and/or fission of lysosomes. Upon treatment with vacuolin-1 (a PIKfyve inhibitor promoting lysosome fusion)²⁶, both control and NKG7-KO CD8⁺ T cells presented enlarged lysosomes with increased diameter, but only control cells effectively recovered their original lysosome size following removal of the drug (Fig. 1g). Conversely, treatment with acetate Ringer’s solution (AR) leads to lysosome fission, substantially reducing the diameter of lysosomes (Fig. 1h). However, upon AR removal, whereas lysosomes of NKG7-KO T cells rapidly fused, the diameter of lysosomes in control cells returned to their basal state at a slower rate. These results suggest that NKG7 regulates lysosomal homeostasis by either slowing/preventing lysosome fusion or accelerating lysosome fission.

Fig. 1. NKG7 regulates homeostasis between late endosomes (LEs) and lysosomes.

a, b Size and number of LAMP1⁺ lysosomes were examined via confocal microscopy in expanded human CD8⁺ T cells treated with indicated Cas9/RNP. a Representative images. b Quantification of (a) in total lysosome numbers (top) and the relative proportion of aggregated lysosomes [cluster] compared to individual lysosomes [spot] (bottom). c Quantification of lysosomes as in (b) for primary human NK cells treated with indicated Cas9/RNP (Supplementary Fig. 1a). d, e Similar to a–c using Jurkat human CD4⁺ T cell line transfected as indicated. f Quantification of lysosomes as in (e) for HEK293T (293T) cells transfected as indicated (Supplementary Fig. 1b). g Expanded human CD8⁺ T cells transfected with indicated Cas9/RNP were treated with vacuolin-1 (10 μM, 37°C, 1 hour) and washed. Upon wash, average lysosome diameter was measured at indicated time points. ‘X’: untreated cells. h Similar to (g) except that CD8⁺ T cells were treated with acetate Ringer’s solution. i A cartoon describing key molecules mediating LE-lysosome fusion process. 41: VPS41, https://BioRender.com/t95p627. j Distribution of Rab7 and Arl8b in expanded human CD8⁺ T cells treated with indicated Cas9/RNP. k Spatial correlation between Rab7 and Arl8b in (j) were quantified. (l) Similar to (k) for VPS41 and LAMP1 (Supplementary Fig. 2e). (a, d, j) White lines show cell boundaries. Scale bar = 3 μm (a, j) or 5 μm (d). (b, c, e, f, k, l) Each faint dot represents a single cell, and dots are color-coded based on the independent experiments. The mean ± SD and statistics based on biological replicates (average from each experiment; bold dots) are superimposed on top. gNeg: non-specific guide RNA, gNKG7: NKG7-targeting guide RNA. The results presented are representative or collected from 3 independent experiments. Statistical analyses were performed using two-tailed paired t-test (b, c, e, f, k, l). Source data are provided as a Source Data file.

Lysosomes undergo dynamic homotypic and heterotypic membrane fusion and fission with other organelles²⁷. Among them, interactions between late endosomes (LEs; represented by Rab7⁺ organelles) and lysosomes (Arl8b⁺ organelles) are critical for lysosomal maturation and macromolecule degradation, and require proteins including Rab7, Arl8b, and the Homotypic fusion and Protein Sorting (HOPS) complex (Fig. 1i)^28,29. In NKG7-KO CD8⁺ T cells, we observed increased association between Rab7⁺ and Arl8b⁺ organelles, suggesting that in the absence of NKG7 there is increased fusion creating hybrid organelles containing markers of both LE and lysosomes (Fig. 1j, k and Supplementary Fig. 2a–d). NKG7-KO cells also showed increased association of the HOPS complex subunit VPS41 with LAMP1⁺ (LE/lysosome) organelles (Supplementary Fig. 2e and Fig. 1l). Taken together, our data suggest that NKG7 coordinates homeostasis between LEs and lysosomes impacting overall lysosomal size and numbers within a cell.

NKG7 regulates mTORC1-mediated metabolic activities in CD8⁺ T cells

While once regarded as a degradative organelle responsible for macromolecular degradation and recycling, the lysosome is now recognized as a central metabolic hub coordinating nutrient sensing and metabolic signaling^25,27,28. To determine whether NKG7 mediates any metabolic process, we first examined the overall metabolic profile of NKG7-KO CD8⁺ T cells. Interestingly, pre-activated NKG7-KO CD8⁺ T cells showed increased glycolysis as well as total glycolytic capacity measured by extracellular acidification rate (ECAR), while mitochondrial respiration of the same cells (represented by the oxygen consumption rate [OCR]), was not significantly affected (Fig. 2a, b and Supplementary Fig. 3a, b). One of the major drivers of glycolytic metabolism is mTORC1, whose activity is essential for the metabolic shift into aerobic glycolysis during TCR-induced activation and proliferation^12,16. Activity of mTORC1 (represented by phosphorylation of S6 [p-S6; S235/36]) in naïve (CD62L⁺CD44^-) and virtual memory (TVM; CD44⁺CD122⁺CD62L⁺) CD8⁺ T cells from CD8-specific Nkg7-KO (CD8-Cre Nkg7^fl/fl) mouse was comparable to that from WT mouse (Supplementary Fig. 3c, d). In addition, NKG7-KO human CD8⁺ T cells presented normal mTORC1 activation upon acute TCR stimulation over a 30-minute time course (Supplementary Fig. 3e, f). However, during chronic T cell activation (from 2 to 7 days following anti-CD3/CD28 stimulation in our experimental setup), NKG7-KO CD8⁺ T cells presented enhanced mTORC1 activity, which was suppressed in the presence of rapamycin (Fig. 2c and Supplementary Fig. 3g, h). Of note, the same cells showed normal phosphorylation of AKT, which is upstream mTORC1 signaling molecule (Supplementary Fig. 3i). Similarly, increased mTORC1 activity during T cell activation was also observed in antigen-primed Nkg7-KO mouse CD8⁺ T cells as well as in human CD8⁺ T cell populations without NKG7 expression compared to their matching controls (Fig. 2d, e and Supplementary Fig. 3j). Furthermore, effector CD8⁺ T cells (CD44⁺CD62L^-) from CD8-specific Nkg7-KO mice presented enhanced mTORC1 activity (Supplementary Fig. 3c, d). These observations suggest that NKG7 restrains mTORC1 activity during chronic T cell activation. In line with augmented mTORC1 activity, NKG7-KO CD8⁺ T cells present enhanced uptake of the glucose analog (2-NBDG) and increased cell size (represented by FSC), at day 4 and 6 post TCR-induced activation (Fig. 2f, g). Overall, our results suggest that NKG7 limits mTORC1-mediated activation and metabolism in CD8⁺ T cells.

Fig. 2. NKG7 negatively regulates mTORC1-mediated metabolic activities.

a Extracellular acidification rate (ECAR) of expanded human CD8⁺ T cells transfected with indicated Cas9/RNP. Oligo: oligomycin, 2DG: 2-deoxyglucose. b Quantification of glycolysis and glycolytic capacity from (a). c Freshly isolated human CD8⁺ T cells were treated with indicated Cas9/RNP and activated with anti-CD3 and anti-CD28 antibodies immobilized on culture plate (pre-coated at 1 μg/mL) with or without rapamycin (Rapa). Cells were harvested at the indicated time point and prepared for flow cytometry of phosphorylated S6 (p-S6; S240/44). Each colored dot indicates an independent experiment from a different blood draw. d Splenic CD8⁺ T cells from control (Ctrl) and CD8cre-Nkg7 KO mice immunized with OVA/poly I:C for 7 days were isolated and p-S6 (S235/36) was examined. [Left] Representative p-S6 histogram among the indicated CD8⁺ T cell populations according to their staining with OVA-specific H-2K^b tetramer (Tet). Numbers represent percent positive population. [Right] Fold change in the percentage of p-S6⁺ among Tet⁺ populations compared to Tet^- populations. e Peripheral blood mononuclear cells (PBMCs) from healthy donors were either rested (top) or treated with anti-CD3 and anti-CD28 antibodies immobilized on beads for 24 hours. Percent positive population of p-S6 (S235/36) was then examined in the gated NKG7⁺/NKG7^- CD8⁺ T cells. [Left] Representative contour plot as indicated. [Right] Percentage of p-S6⁺ among the indicated CD8⁺ T cell populations. f Freshly isolated human CD8⁺ T cells were treated with indicated Cas9/RNP and activated as in (c). Cells were collected at indicated time points and labelled with 2-NBDG (fluorescent glucose analog). 2-NBDG fluorescence and forward scatter (FSC; cell size) were measured via flow cytometry. g Quantifications of percent positive 2-NBDG^hi and FSC^hi in (f). gNeg: non-specific guide RNA, gNKG7: NKG7-targeting guide RNA. The results presented are representative or collected (mean ± SD) from 5 (c), 4 (e), 2 (d), and 3 (rest) independent experiments. Statistical analyses were performed using two-tailed paired t-test (b, g), two-tailed unpaired t-test (d, e), or one-way ANOVA with Dunnett’s multiple comparisons (c). Source data are provided as a Source Data file.

NKG7 inhibits mTORC1 translocation to lysosomes

Activation of mTORC1 is tightly coordinated via 2 major regulatory mechanisms: 1) nutrient-induced mTORC1 recruitment to lysosomes and 2) signaling cascades induced by environmental stimuli that lead to activation of RHEB GTPases, which activate mTOR on the lysosome^10,11. Since we observed increased mTORC1 activation upon chronic TCR stimulation in NKG7-KO CD8⁺ T cells, we examined whether NKG7 might affect mTOR recruitment to lysosomes. Indeed, in NKG7-KO CD8⁺ T cells, a larger proportion of mTOR was found at the enlarged lysosomes compared to that in control cells (Fig. 3a, b). Conversely, ectopic NKG7 expression in Jurkat cells decreased the frequency of lysosome-associated mTOR (Fig. 3c, d). Amino acids are essential nutrients required for mTORC1 activation and increase in amino acids has been shown to mediate mTORC1 translocation to lysosomes and subsequent activation^10,11,30. To examine whether NKG7 regulates amino acid-mediated mTOR localization to lysosomes, we co-plated 293 T cells stably expressing either YFP or YFP-NKG7 and examined mTOR distribution within each cell type, during amino acid-starvation and -stimulation (Fig. 3e, f). In the case of YFP-expressing cells, mTOR is dispersed in the cytoplasm upon amino acid starvation (Fig. 3e, inset 1), but is rapidly recruited to lysosomes upon amino acid repletion (Fig. 3e, inset 3) as previously described³⁰. However, cells with ectopic YFP-NKG7 expression show diminished mTOR recruitment to the lysosome upon amino acid addition (Fig. 3e [insets 2 and 4], f). Importantly, this impaired mTOR recruitment to the lysosomes in YFP-NKG7-expressing cells led to delayed amino acid-induced mTORC1 activation as measured by the phosphorylation of several downstream targets (Fig. 3g). Altogether, our results suggest that NKG7 negatively regulates amino acid-mediated mTOR translocation to the lysosome and subsequent activation.

Fig. 3. NKG7 negatively regulates mTOR translocation to lysosomes.

a Intracellular mTOR distribution relevant to lysosomes (LAMP1⁺) in expanded human CD8⁺ T cells treated with indicated Cas9/RNP. gNeg: non-specific guide RNA, gNKG7: NKG7-targeting guide RNA. b Spatial correlation between mTOR and LAMP1 in (a). c, d Similar to (a and b) using Jurkat T cells transfected as indicated. e 293 T cells stably expressing YFP or YFP-NKG7 were plated together and rested in amino acid (a.a.)-free medium for 50 min (-a.a.). Amino acids were then added to the rested cells for 10 min ( + a.a.), and the cells were then prepared for microscopy staining with the indicated antibodies. Regions designated by the white square with number are magnified on the right. f Spatial correlation between mTOR and LAMP1 in (e). g 293 T cells stably expressing YFP or YFP-NKG7 were starved of a.a. as in (e) and stimulated with a.a. for indicated time points. Samples were then prepared for immunoblot and phosphorylation and expression of the indicated proteins were examined. (a, c, e) Scale bar = 3 μm (a), 5 μm (c), or 10 μm (e). White (a, c) and yellow (e) lines show boundaries of a cell and YFP-NKG7-expressing cells, respectively. (b, d, f) Each faint dot represents a single cell, and dots are color-coded based on the independent experiments. The mean ± SD and statistics based on biological replicates (average from each experiment; bold dots) are superimposed on top. The results presented are representative or collected from 3 independent experiments. Statistical analyses were performed using two-tailed paired t-test (b, d) or one-way ANOVA with Dunnett’s multiple comparisons for (f). Source data are provided as a Source Data file.

NKG7 inhibits v-ATPase-dependent lysosomal acidity and mTORC1 regulation

To determine how NKG7 impacts mTOR recruitment to lysosomes, we sought to identify NKG7-interacting proteins by 1) NKG7-immunoprecipitation (IP) from NKG7-overexpressing lysates followed by mass-spectrometry and 2) NKG7-proximity labeling via TurboID followed by mass-spectrometry (Supplementary Fig. 4a, b). We also included public NKG7-Y2H database³¹ in our analyses and screened for proteins expressed in human CD8⁺ T cells³², considering the CTL-specific expression of NKG7²⁰. While we did not note any protein that was specific to CD8⁺ T cells, it is well appreciated that mTORC1 translocation to lysosomes is mediated by binding of mTORC1 to the heterodimeric Rag GTPases, which are tethered to the Ragulator complex. In addition, the vacuolar H⁺ ATPase (v-ATPase), which localized at the lysosome and constitutively interacts with the Ragulator complex, is critical for mTORC1 recruitment^10,11,33. Interestingly, our proteomic data identified RagC (Rag GTPase subunit) and LAMTOR1 (Ragulator subunit) as well as v-ATPase components ATP6V0d1 and ATP6AP2 as potential NKG7-interacting proteins (Fig. 4a, b and Supplementary Data 1-2). We initially sought to confirm the interactions of NKG7 with ATP6AP2 and ATP6V0d1 via co-immunoprecipitation from lysates of cells overexpressing NKG7 along with ATP6AP2 (Fig. 4c) and ATP6V0d1 (Fig. 4d), respectively. In the case of ATP6V0d1, we further validated the interaction with NKG7 via co-IP of endogenously expressed proteins in expanded CD8⁺ T cells (Fig. 4e).

Fig. 4. NKG7 inhibits v-ATPase-mediated lysosomal homeostasis and mTORC1 regulation.

a Venn diagram highlights of candidate NKG7-interacting proteins involved in mTORC1 regulation. https://BioRender.com/e43h618b Amino acid-induced mTORC1 recruitment to lysosomes. Red: proteins shown in (a). d1: ATP6V0d1. https://BioRender.com/f92t271c, d FLAG (F) was immunoprecipitated (IP) from 293 T cell lysates with indicated transfections and immunoblotted. N7: NKG7, AP2: ATP6AP2, V0d1: ATP6V0d1, mCh: mCherry. e NKG7 IP from human CD8⁺ T cell lysates. rIgG: rabbit IgG. f Relative lysosomal acidity of indicated human CD8⁺ T cells measured by LysoSensor dye. g Lysosomal pH of YTS cells with indicated conditions. Each colored dot indicates an independent experiment. KO: NKG7-KO clone. h Similar to (f) using Jurkat T cells with NKG7 overexpression (OE). Empty: control plasmid. i Similar to (g) with indicated 293 T cells. j Cathepsin D (CatD) processing of indicated human CD8⁺ T cells. (top) Red arrow: pre-processed form, red arrowhead: mature form. C: gNeg, N7: gNKG7. (bottom) Quantifications of (top). Relative CatD processing (mature / pre-processed) was normalized to the level of gNeg group. k Similar to (j) using indicated 293 T cells. l mTOR distribution relevant to lysosomes (LAMP1⁺) of indicated human CD8⁺ T cells treated with BafA1. m Spatial correlation between mTOR and LAMP1 in (l). n, o Similar to (l, m) except that indicated human CD8⁺ T cells were further transfected with gATP6AP2 Cas9/RNP. l, n White line: a boundary of a cell. Scale bar = 3 μm. m, o Each faint dot represents a single cell, and dots are color-coded based on the independent experiments. The mean ± SD and statistics based on biological replicates (average from each experiment; bold dots) are superimposed on top. gNeg: non-specific guide RNA, gNKG7/gATP6AP2: guide RNA targeting NKG7/ATP6AP2 respectively. The results presented are representative or collected (mean ± SD) from 5 (c, g), 4 (d, e), or 3 (f, h–o) independent experiments. Statistical analyses were performed using two-tailed one-sample t-test with the hypothetical value set as 1 (j, k) or one-way ANOVA with Dunnett’s (g, i, o) or Tukey’s (m) multiple comparisons. Source data are provided as a Source Data file.

The action of v-ATPase is critical for the maintenance of acidic lysosomal pH and is required for optimal processing of hydrolytic enzymes and their degradative activities^34,35. In human and mouse CD8⁺ T cells as well as in the human NK cell line YTS, the absence of NKG7 caused an increase in lysosomal acidity (Fig. 4f, g and Supplementary Fig. 4c, d), whereas ectopic NKG7 expression in Jurkat and 293 T cells led to more basic lysosomes (Fig. 4h, i). In line with these observations, cathepsin D (lysosomal aspartic protease) was mainly detected as a fully processed mature form in NKG7-KO CD8⁺ T cells compared to the control cells (Fig. 4j)³⁶. However, an immature form of cathepsin D (procathepsin D) was prominent in 293 T cells overexpressing NKG7 (Fig. 4k). These observations indicated that NKG7 negatively regulates v-ATPase-dependent lysosomal acidity. Indeed, inhibition of v-ATPase activity with Bafilomycin A1 (BafA1) increased lysosomal pH in YTS and Jurkat cells, and the pH levels were comparable whether or not NKG7 was present (Fig. 4g, i). We also examined whether NKG7-mediated negative regulation of mTORC1 is via v-ATPase activity. Indeed, BafA1 treatment of NKG7-KO CD8⁺ T cells resulted in the cytosolic redistribution of mTOR (Fig. 4l, m). Additionally, deletion of ATP6AP2, which affects v-ATPase membrane assembly and activity^37,38, resulted in mTOR cytosolic redistribution in NKG7-KO CD8⁺ T cells (Fig. 4n, o and Supplementary Fig. 4e, f). We further found that BafA1 treatment of 293 T cells delayed amino acid-induced mTORC1 activation in a dose-dependent manner, phenocopying NKG7-expressing cells (Supplementary Fig. S4g, h and Fig. 3g). Taken together, our results indicate that NKG7 inhibits v-ATPase-dependent lysosomal activity and mTORC1 recruitment to lysosomes.

NKG7 negatively regulates v-ATPase activity by inhibiting the association between v-ATPase V0-V1 domains

The hetero-multimeric v-ATPase complex is composed of 13 subunits, contained within a membrane-embedded V0 domain, which functions as a proton pump, and a cytosolic V1 domain, which provides energy required via ATP hydrolysis (Fig. 4b)^34,35. The primary mechanism regulating v-ATPase activity is the reversible assembly and disassembly of the cytosolic V1 and membrane-bound V0 domains. To investigate whether NKG7 affects the association of V1 and V0 domains, we examined the lysosomal level of V1 domain subunits in 293 T cells expressing NKG7 compared to cells without NKG7 (Fig. 5a). LAMP1 and GAPDH were detected only in the crude lysosomal fraction (CLF) and in non-lysosomal fraction (NLF), respectively, matching with their expected distribution. Interestingly, we observed a reduced level of v-ATPase V1 domain subunits (ATP6V1A1 and ATP6V1E1) in lysosomes of NKG7-expressing cells (Fig. 5a, b [top]). Of note, we observed increased levels of V0 subunits (ATP6V0a1 and ATP6V0d1) and an accessory subunit (ATP6AP2) in NKG7-expressing lysosomes (Fig. 5a, b [bottom]), which might be due to the interaction of NKG7 with the V0 domain (Fig. 4c–e). We also examined the impact of NKG7 on V0-V1 association by performing the proximity ligation assay (PLA) between V0 subunit (ATP6V0a4) and V1 subunit (ATP6V1B1)³⁹. PLA detects the presence of two proteins within a 40 nm range; therefore, PLA puncta represent assembly of V0-V1 domains. Ectopic expression of NKG7 in 293 T cells reduced the number of PLA puncta within a cell, whereas NKG7 depletion in YTS NK cells increased the number of PLA puncta (Fig. 5c–f and Supplementary Fig. 5a), suggesting that NKG7 interferes with V0-V1 association leading to a functional membrane-bound v-ATPases. We also examined V1-V0 domain assembly on membrane fractions as previously described⁴⁰. NKG7 expression reduced the presence of V1 domain subunits in the membrane fraction supporting that NKG7 expression impairs the recruitment of the cytosolic V1 complex to the membrane bound V0 domain (Supplementary Fig. 5b, c). Conversely, we observed increased membrane levels of V1 domain subunits in NKG7-KO CD8⁺ T cells (Supplementary Fig. 5d, e). Collectively, our data suggest that NKG7 impairs v-ATPase activity by preventing membrane assembly of the V1 and V0 domains.

Fig. 5. NKG7 inhibits association between v-ATPase V0-V1 domains and interrupts interaction between Ragulator and v-ATPase complexes.

a A crude lysosomal fraction (CLF) from 293 T cells expressing indicated proteins. Relative protein levels within the CLF were examined. PNL post-nuclear lysate, NLF non-lysosome fraction. b Quantification of (a). Densitometry of indicated protein was normalized to LAMP1. c, d Proximity ligation assay (PLA) between ATP6V0a4 (V0a4) and ATP6V1B1 (V1B1) using 293 T cells expressing indicated proteins (c), and its quantification (d). e, f Similar to (c, d) using indicated YTS cells. g LAMTOR1 distribution relevant to lysosomes (LAMP1⁺) in indicated human CD8⁺ T cells. h Spatial correlation between LAMTOR1 and LAMP1 in (g). i, j Similar to (g, h) using Jurkat T cells transfected as indicated. k HA was immunoprecipitated (IP) from lysates of 293 T cells expressing indicated proteins as well as transfected with LAMTOR1(LT1)-HA plasmid and immunoblotted. l Quantification of (k). m Indicated human CD8⁺ T cells were treated with DMSO or bafilomycin A1 (BafA1), and LAMTOR1 distribution relevant to lysosomes was examined. n Spatial correlation between LAMTOR1 and LAMP1 in (m). a, k YFP: FLAG-YFP, N7: FLAG-YFP-NKG7. (b, l) V1A1: ATP6V1A1, V1E1: ATP6V1E1, V0d1: ATP6V0d1, V0a1: ATP6V0a1, AP2: ATP6AP2, LT3: LAMTOR3. Each dot represents quantification from an individual experiment. RU: relative unit. c, e, g, i, m Yellow/white line: a boundary of a cell. Scale bar = 3 (g, m), 5 (e, i), or 10 (c) μm. (d, f, h, j, n) Each faint dot represents a single cell, and dots are color-coded based on the independent experiments. The mean ± SD and statistics based on biological replicates (average from each experiment; bold dots) are superimposed on top. gNeg: non-specific guide RNA, gNKG7: NKG7-targeting guide RNA. The results presented are representative or collected (mean ± SD) from 3 independent experiments. Statistical analyses were performed using two-tailed one-sample t-test with hypothetical value set as 1 (b, l), two-tailed paired t-test (d, f, h, j), or one-way ANOVA with Dunnett’s multiple comparisons (n). Source data are provided as a Source Data file.

NKG7 impairs Ragulator complex localization to lysosomes and disrupts the association between the Ragulator and v-ATPase complexes

The v-ATPase complex is constitutively associated with the Ragulator complex via direct interactions between ATP6V0d1 and LAMTOR1³³. In this regard, it was interesting to note that LAMTOR1 was also identified as a potential NKG7-interacting protein in our proximity-dependent biotinylation experiments (Fig. 4a and Supplementary Data 1-2). Therefore, we investigated if NKG7 might disrupt the association between the v-ATPase and Ragulator complexes. LAMTOR1 anchors the Ragulator complex to the lysosomal membrane via its N-terminal myristoylation and palmitoylation⁴¹. In normal CD8⁺ T cells, the distribution of LAMTOR1 and LAMP1 correlate with each other, but LAMTOR1 was found to be localized in a specific compartment within LAMP1⁺ organelles using Airyscan high resolution confocal imaging (Fig. 5g [left], h). However, upon NKG7-KO, that unique localization of LAMTOR1 with the lysosome was not observed and LAMTOR1 and LAMP1 distributions were found to significantly overlap (Fig. 5g [right], h). Interestingly, in cells which do not express endogenous NKG7 (Jurkat and 293 T cells), LAMTOR1 and LAMP1 distributions overlaid each other similar to NKG7-KO cells (Fig. 5i [left], j and Supplementary Fig. 5f [insets 1], g). However, upon ectopic expression of NKG7, LAMTOR1 presented the unique lysosome distribution similar to normal CD8⁺ T cells (Fig. 5g–j and Supplementary Fig. 5f [inset 2], g). We next tested whether NKG7 affects the interaction between the v-ATPase and the Ragulator complexes. Although NKG7 expression did not affect the interaction of LAMTOR1 with the other Ragulator complex subunit (LAMTOR3) or Rag GTPase (RagA), the levels of v-ATPase subunits associated with LAMTOR1 was reduced (Fig. 5k, l). We also examined whether the unique localization of the Ragulator complex with lysosomes is dependent on v-ATPase activity. Indeed, BafA1 treatment of NKG7-KO CD8⁺ T cells resulted in LAMTOR1 distribution similar to that observed in control CD8⁺ T cells (Fig. 5m, n). Altogether, our results suggest that NKG7 impacts Ragulator complex recruitment to the v-ATPase-residing lysosomal compartment and subsequent activation of mTORC1 by impairing V1 and V0 v-ATPase assembly on the lysosomal membrane.

Nkg7 promotes durable CD8⁺ T cell responses to LCMV infection

T cell differentiation into memory populations requires metabolic rewiring into a quiescent mode, mainly relying on oxidative phosphorylation (OXPHOS) and fatty acid oxidization (FAO)^12,16,42. In this regard, sustained mTORC1 activity of T cells resulted in active glycolytic metabolism promoting potent effector functions, while impairing conversion into memory populations^43,44. On the other hand, interference of mTORC1 activity promotes generation of memory T cells, suggesting that regulation of mTORC1 activity plays a critical role in determining T cell fate^43,45,46. To explore physiological roles of NKG7-mediated mTORC1 regulation, we assessed if Nkg7 regulates in vivo CD8⁺ T cell responses to viral infection. We first knocked out Nkg7 in LCMV-specific P14 CD8⁺ T cells via Cas9/RNP nucleofection and co-transferred with control P14 cells into congenically distinct mice, which were subsequently infected with LCMV virus (Fig. 6a). In response to either acute (LCMV-Arm) or chronic (LCMV-Clone 13; LCMV-Cl13) infection, Nkg7-KO (sgNkg7) P14 cells normally expanded at the first week after infection, but their numbers precipitously declined after this period (Fig. 6b). In response to LCMV-Arm, Nkg7-KO P14 cells had decreased proportions of the CD62L⁺ central memory subset (Fig. 6c), as well as decreased percentages of memory precursor cells at the peak of acute phase of infection (Fig. 6d, e). Conversely, KLRG1⁺ terminal effector CD8⁺ T cells were more abundant in Nkg7-KO (Fig. 6d, e). Nkg7-KO P14 cells also displayed aberrantly high expression of KLRG1 in response to LCMV-Cl13 (Fig. 6f, g), which is associated with a terminal phenotype and shorter lifespan in chronic LCMV⁴⁷. At the chronic phase of infection (day 30), not only the total numbers of Nkg7-KO P14 cells were lower but the accumulation of stem-like Ly108⁺ P14 cells was significantly affected (Fig. 6h, i). We found similar results when tracking endogenous LCMV-specific CD8⁺ T cell responses in WT and CD8-specific Nkg7-KO (CD8-Cre Nkg7^fl/fl) mice: decreased proportions of memory precursors and increased amounts of terminal effectors in response to LCMV-Arm (Supplementary Fig. 6a, b), and decreased numbers of LCMV-specific CD8⁺ T cells after 30 days of infection (Supplementary Fig. 6c). In addition, LCMV-Cl13 infected CD8-specific Nkg7-KO mice had increased kidney viral titers compared to WT mice (Fig. 6j). In line with our earlier observations, enhanced mTORC1 activity (represented by p-S6 [S235/36]) was observed in Nkg7-KO P14 cells upon infection with both LCMV-Arm and LCMV-Cl13 (Fig. 6k, l), and the same was observed for endogenous LCMV-specific Nkg7-KO CD8⁺ T cells (Supplementary Fig. 6d). In addition, blockade of v-ATPase activity via in vivo BafA1 treatment reverted enhanced mTORC1 activity in Nkg7-KO P14 cells similar to what we observed in human CD8⁺ T cells (Figs. 6k, l and 4l, m). Importantly, these changes were reflected in the fate of Nkg7-KO P14 cells, which were significantly rescued with low-dose rapamycin-based mTORC1 inhibition – especially T_CM cells (Supplementary Fig. 6e–i) or BafA1 treatment (Fig. 6m). Finally, CRISPR-Cas9-mediated knockout of Atp6v0d1 significantly rescued the memory accumulation of Nkg7-KO P14 cells (Fig. 6n–p). Altogether, our results suggest that Nkg7 is necessary for the long-term establishment of CD8⁺ T cell responses to viral infections, through negative regulation of mTORC1.

Fig. 6. NKG7 promotes CD8⁺ T cell durability and function in response to viral infections.

a–i Naïve P14 cells were transfected with a mix of Cas9:sgRNAs for Cd19 (WT) or Nkg7 (KO) as indicated. WT and KO cells (1:1 mix) were transferred into recipient mice infected with LCMV-Arm or LCMV-Cl13. a Experimental schematics. https://BioRender.com/h76z412b KO/WT cell ratios over time. Dashed lines: transfer ratios. c Percentages of central memory (CD44⁺CD62L⁺) P14 cells (LCMV-Arm, day 30-infected). Connecting lines: co-transferred pairs. d Flow cytometry showing terminal effector (TE; KLRG1⁺CD127^-) and memory progenitor (MP; KLRG1^-CD127⁺) P14 cells upon indicated condition. D: day post-infection. e Percentages of TE and MP P14 cells upon indicated conditions (n = 5). f Flow cytometry showing KLRG1 in P14 cells upon indicated condition. Gray: naïve CD8⁺ T cells (control). g KLRG1 gMFI of (f). n = 5. h Flow cytometry showing Tim3 and Ly108 in P14 cells upon indicated condition. i Percentages of Ly108⁺ P14 cells as in (h). Connecting lines: co-transferred pairs. j LCMV PFU values for WT (Nkg7^fl/fl; n = 10) or Nkg7-KO (Cd8-Cre Nkg7^fl/fl; n = 7) mice infected with LCMV-Cl13. k–m Similar to (a), where mice were treated with vehicle/BafA1 between days 1-3 post-infection. k, l p-S6 gMFI of P14 cells upon indicated conditions. n = 8 (k) and 5 (l). m KO/WT P14 cell ratios (LCMV-Arm, day 14-infected). n = 5 (vehicle) and 6 (BafA1). n–p Naïve P14 cells from WT (Nkg7^fl/fl) or Nkg7-KO (Cd8-Cre Nkg7^fl/fl) mice were treated with control (sgCtrl; sgCd19) or sgAtp6v0d1 Cas9/RNP and transferred to LCMV-Arm-infected mice. n Flow cytometry showing CD62L (T_CM cells) in the indicated groups. o Nkg7-KO/WT P14 cell ratios (n = 4). p Nkg7-KO/WT P14 T_CM cell ratios (n = 4). The results presented are representative or collected (mean ± SD) from 3 (j) and 2 (rest) independent experiments. Statistical analyses were performed using two-tailed multiple t-tests (comparison to transfer ratios) for (b), two-tailed paired t-tests (c, i), one-way ANOVA with Tukey’s multiple comparisons (e, k, l), or two-tailed unpaired t-test (g, j, m, o, p). Source data are provided as a Source Data file.

CD8⁺ T cell specific NKG7 overexpression leads to robust anti-tumor immunity

We previously demonstrated that NKG7 was upregulated in the circulating effector/memory CD8⁺ T cells of the responders to immune checkpoint inhibitor (ICI) therapy compared to non-responders¹⁹. To investigate whether NKG7 upregulation in CD8⁺ T cells is sufficient for their efficient anti-tumor responses, we created CD8⁺ T cell specific NKG7 transgenic (NKG7 Tg) mice (Fig. 7a). In this model, we first produced the pCAG2-NKG7-Rosa26-IRES-tdTomato (referred to as “Rosa26-NKG7” from here) mice and crossed them with CD8 (E8I)-Cre mice to generate CD8-Cre-NKG7 transgenic mice, in which Cre will delete the stop codon before NKG7 and induce enhanced NKG7 expression in CD8⁺ T cells. Bicistronic expression of tdTomato was used as a proxy marker to identify the NKG7 overexpression in CD8⁺ T cells (Fig. 7b). From there, we evaluated how CD8⁺ T cells responded to immunogenic tumors in NKG7 Tg mice upon subcutaneous injection with MC38 colon adenocarcinoma tumor cells. We found that tumor growth was significantly delayed in NKG7 Tg mice compared to control mice (Rosa26-NKG7 without CD8-Cre; Tg Cre-), suggesting that NKG7 promotes CD8⁺ T cell anti-tumor responses (Fig. 7c). At the endpoint (day 17 after tumor injection), tumor infiltrating lymphocytes (TILs) and spleen cells from the tumor-bearing NKG7 Tg and control mice were isolated for functional analysis. Overall accumulation of CD8⁺ T cells within the tumor was greater in NKG7 Tg mice compared to control mice (Fig. 7d). We also observed a greater population of functional (IFN-γ⁺ and CD107a ⁺) CD8⁺ T cells among TILs in NKG7 Tg mice, but the expression of perforin and granzyme B were comparable between control and NKG7 Tg mice in the TILs and spleen (Fig. 7e–g and Supplementary Fig. 7a, b), suggesting that NKG7 overexpression can enhance the delivery or trafficking of lytic granules but does not affect the contents of lytic granules of tumor-reactive CD8⁺ T cells within tumor tissues. Taken together, our results indicate that overexpression of NKG7 in CD8⁺ T cells promotes in vivo anti-tumor activities.

Fig. 7. CD8⁺ T cell specific NKG7 overexpression enhances anti-tumor immunity and NKG7 expression level is associated with survival of patients with aggressive bladder cancer.

a Production of CD8⁺ T cell specific NKG7 transgenic (Tg) mice. b Validation of NKG7 Tg mice with reporter gene (tdTomato) expression in splenic CD8⁺ T cells. c–g 5 ×10⁵ MC38 tumor cells were injected subcutaneously at the right flank, the tumor size was measured twice a week till to the humane endpoint, and the lymphocytes were isolated from tumor tissues and spleen on day 14-16 after tumor injection. c Tumor growth (top) and survival curve (bottom) of control (Tg Cre-; n = 10) and NKG7 Tg (n = 8) mice. d Frequency of CD8⁺ T cells in tumor infiltrating lymphocytes (TILs; n = 9) and spleen (n = 13 and 10). e Representative flow cytometry plots showing IFN-γ producing CD8⁺ T cells in TILs (n = 7 and 9) and spleen (n = 10 and 9). f Quantification of IFN-γ producing CD8⁺ T cells in (e). g Quantification of CD8⁺ T cell degranulation (CD107a⁺) in TILs (n = 9 and 12). h–j NKG7 expression levels were measured by tissue staining of resected tumors from patients with bladder cancer (n = 158) and prognostic value was determined on overall survival. h Representative images of NKG7 staining in cystectomy specimens of a patient with negative expression of NKG7 and a patient with positive expression of NKG7. Scale bar – 100 μm i Kaplan-Meier curves for overall survival (OS) of patients undergoing radical cystectomy for localized bladder cancer and stratified with NKG7 expression. j Univariate and multivariate analysis of the risk of death based on NKG7 mid/high expression versus NKG7 neg/low. The results presented are representative or collected (mean ± SD) from 2 (b–g) independent experiments. Statistical analyses were performed using two-tailed unpaired t-test (d, f, g) and Two-way ANOVA (P < 0.001) in tumor growth. Log-rank test (P < 0.05) of survival were performed in (c). Cox proportional hazard regression models and presented using hazard ratios (HR), 95% confidence intervals (CI), and one-tailed p value. Source data are provided as a Source Data file.

Elevated NKG7 expression is associated with prolonged overall survival of patients with bladder cancer

To determine the prognostic value of NKG7 expression in cancer, we first analyzed overall survival outcomes of patients with low and high NKG7 gene expression in TCGA datasets (Supplementary Fig. 7c). Among 16 cancers tested, 11 showed significant association of NKG7 gene expression with overall survival. Only 1 cancer type (Kidney Renal Papillary Cell Carcinoma [KIRP]) showed poor outcome for patients with high NKG7 expression. For 10/11 cancers, high NKG7 expression was associated with good prognosis. Bladder cancer is characterized by a high rate of recurrence post-surgery in localized disease and intrinsic resistance to immunotherapy is common. Based on the critical role of NKG7 in CD8⁺ T cell function, we hypothesize that NKG7 expression can be a prognostic marker of survival outcome. To test this, we analyzed the association of NKG7 gene expression in the TCGA bladder cancer dataset (BLCA). Prolonged overall survival was found in patients with high NKG7 expression (HR: 0.62, 95% CI: 0.44-0.88, p = 0.0076) (Supplementary Fig. 7d). To validate these findings, we scored NKG7 protein expression in tumor-infiltrating T cells from a cohort of 158 patients diagnosed with muscle-invasive bladder cancer and treated with radical cystectomy (Fig. 7h). Immunofluorescence staining of cystectomy specimens confirmed the presence of CD8⁺ T cells with negative/low and medium/high NKG7 expression in intracellular granules (Supplementary Fig. 7e). High NKG7 expression was significantly associated with longer overall survival (Fig. 7i) with a median survival time of 18.9 years compared to 2.65 years for patients with low NKG7 expression (p < 0.0001). By univariate analysis, high NKG7 expression was significantly associated with a reduced risk of death (HR:0.49, 95% CI: 0.34-0.73, p < 0.0001) (Fig. 7j). Similar findings were observed with multivariable analysis by including other risk factors of death such as sex, lymph node positivity, T stage and TIL infiltration. Altogether, these clinical data demonstrate that NKG7 is an independent predictor of overall survival in bladder cancer.

Discussion

NKG7 has been shown to be an important regulator of CD8⁺ T cell-mediated anti-tumor immunity through its effects on lytic granule release^19,22,24. Herein, we demonstrate that NKG7 is not just a regulator of cellular cytotoxicity but is also a critical gatekeeper of mTORC1 activity in CD8⁺ T cells through its perturbation of v-ATPase lysosomal assembly. This activity of NKG7 regulates both CD8⁺ T cell durability and memory generation in response to infection. Moreover, we show that CD8⁺ T cell-specific overexpression of NKG7 leads to increased numbers of tumor-infiltrating lymphocytes, and cytokine expression, resulting in enhanced anti-tumor immunity. Overall, our results identify NKG7 as a cell-type specific modulator of mTORC1 activity impacting CD8⁺ T cell function, durability, and differentiation.

Prior studies have identified that the loss of NKG7 in CD8⁺ T cells and NK cells leads to defective lytic granule release^19,22,24. We now show that loss of NKG7 also results in larger but fewer Lamp1⁺ lysosomes, as a result of aberrant LE/lysosome fusion. In contrast, overexpression of NKG7 in cells that do not express it (Jurkat T cells and 293 T cells) resulted in more numerous, but smaller lysosomes compared to control transfected cells. While it remains to be determined how NKG7 can modulate this process, these hybrid organelles might affect the normal maturation of lytic granules, or even fusion with the plasma membrane, such as that seen in patients with Chediak Higashi Syndrome⁴⁸. Notably, deletion of ATP6AP2 in NKG7-KO cells did not reverse the hybrid organelle phenotype suggesting that increased v-ATPase activity is likely not involved in their formation. Future studies aimed at defining the mechanisms by which NKG7 regulates hybrid organelle generation will be needed to better understand this function of NKG7.

mTORC1 lysosomal recruitment and activity are regulated through the interaction of the Ragulator complex with the assembled v-ATPase. Here, we show that NKG7 interacts with the membrane-integrated V0 domain component ATP6V0d1 and V0 domain-associated accessory subunit ATP6AP2 and inhibits v-ATPase proton pump activity. In the absence of NKG7, v-ATPase activity was dysregulated resulting in increased lysosomal acidity, and enhanced mTORC1 recruitment and activation. In contrast, overexpression of NKG7 in Jurkat T cells or 293 T cells decreases v-ATPase membrane recruitment, activity and Ragulator interaction, impairing lysosomal mTORC1 activation. v-ATPase activity is required for mTORC1 regulation as treatment of cells with the v-ATPase inhibitor, concanamycin A or salicylihalamide A, impairs mTORC1 activity³³. Significantly, treatment with the v-ATPase inhibitor BafA1 or deletion of ATP6AP2 reversed mTORC1 activation in NKG7-KO CD8⁺ T cells. Additionally, hyperactive mTORC1 activity found in Nkg7-KO CD8⁺ T cells from LCMV-infected mice was also reversed with BafA1 treatment. Lastly, the constitutive association of the v-ATPase and Ragulator complex is mediated, among other binding activities, by the interaction between ATP6V0d1 and LAMTOR1³³. Indeed, LAMTOR1 was also identified as a potential NKG7-interacting protein via both our TurboID- and IP-based proteomic approaches. Change in interaction of Ragulator with the V1 domain but not with the V0 domain was observed upon amino acid stimulation³³. However, we found that NKG7 expression impacted Ragulator interactions with both V0 and V1 domains, supporting the notion that NKG7 might directly interrupt the association between v-ATPase and Ragulator complexes in addition to inhibiting v-ATPase assembly and activity.

The diverse nutrient sensing mechanisms regulating mTORC1 recruitment and activation appear to be highly conserved among eukaryotes. To the best of our knowledge, NKG7 represents the first cell type-specific metabolic regulator that enables differential modulation of this ubiquitous metabolic machinery for CD8⁺ T cells. While mTORC1 activity is critical for the expansion and generation of CD8⁺ effector T cells following activation, suppression of mTORC1 signaling is required for memory CD8⁺ T cell generation^43–46. Therefore, understanding mTORC1 regulation by NKG7 is necessary to define the exact mechanisms by which CD8⁺ T cells fine-tune metabolic processes to maintain their high cytolytic activity in nutrient-limited conditions, such as within the tumor tissue or areas of active infection^12,16,42. In addition, it will help determine how mTORC1 activity is regulated during T cell activation, to promote the expansion and function of effector CD8⁺ T cells, without impairing their ability to transition into memory.

We found that Nkg7 promotes the longevity of LCMV-specific CD8⁺ T cells in response to acute or chronic LCMV infection. Nkg7-KO preferentially impacted CD8⁺ T cell subsets with intrinsic long-lived properties, such as CD62L⁺ central memory cells or stem-like exhausted Ly108⁺ cells^49,50. It is not entirely clear whether this is a result of Nkg7 specifically acting on already formed memory-like precursors. There is lower Nkg7 mRNA expression in stem-like exhausted CD8⁺ T cells^51,52. In addition, the effect of Nkg7 in downregulation of mTORC1 activity is already apparent prior to the full development of memory-like populations (Fig. 6). This evidence may suggest Nkg7-mediated mTORC1 downregulation acts to promote a pro-survival fate during the effector phase of CD8⁺ T cell responses. This is also reflected in the transition from effector to memory. There is a precipitous loss of CD8⁺ T cells in the contraction phase of the immune response, which happens beyond percentage losses of memory precursor cells. The reasons why this precipitous loss occurs is still unclear. It is possible that Nkg7-KO-mediated exacerbation of mTORC1 activation leads to apoptotic cell death in both effector and memory precursor CD8⁺ T cells, which could be induced by sustained loss of intracellular ATP and induction of energetic stress⁵³. Alternatively, distinct memory CD8⁺ T cell subsets may be affected by Nkg7-KO in an mTORC1-dependent or independent way, as evidenced by the selective rescue of T_CM cells, but not other subsets, by low-dose rapamycin. Future studies tracking the fate of Nkg7⁺ CD8⁺ T cells at a single-cell level will be important to test, which different pro-survival pathways can be promoted by Nkg7 in different memory CD8⁺ T cell subsets. It will also be important to understand if Nkg7 expression is asymmetrically regulated during early effector CD8⁺ T cell responses like other fate-defining factors^54,55, or if Nkg7 expression modulation occurs uniformly among all activated CD8⁺ T cells. The latter would suggest a role for the Nkg7-v-ATPase-mTORC1 in the dedifferentiation of effector CD8⁺ T cells into memory-like populations⁵⁶.

An effective immunotherapy of cancer depends on a durable supply of effector cytotoxic T cells to attack growing cancer cells; therefore, a significant T cell infiltration in tumor tissues has been used as a proxy indicator of “hot” tumors for clinical responses to immunotherapy. However, not all patients with “hot” tumor signatures are responsive to ICI therapy^57,58. To that end, our results suggest that infiltrating T cells need expression of NKG7 to optimize their anti-tumor activity in harsh tumor microenvironments. Using CD8-specific conditional NKG7 transgenic mouse model, we show that a CD8⁺ T cell-specific overexpression of NKG7 leads to a strong spontaneous anti-tumor immunity in vivo. We observed an increased accumulation of functional CD8⁺ T cells within tumors in CD8-specific NKG7 transgenic mice compared to control mice, suggesting that NKG7 overexpression can harness durable tumor-reactive CD8⁺ T cells within tumor tissues. This is in line with a previous study using conventional NKG7-deficient mice, which identified a necessary role of NKG7 in response to ICI therapy and accumulation of T cells at tumor sites⁵⁹. The results also support our previous clinical observation of a high basal level of NKG7 transcription in tumor tissues of cancer patients who were responsive to ICI therapy¹⁹, which can be explained as the presence or accumulation of more effective effector CD8⁺ T cells within tumors.

Our study of advanced bladder cancer further supports the role of NKG7 in accumulation of effector CD8⁺ T cells in tumor tissues, which is associated with long-term protection and survival of these patients. The clinical implication of NKG7 expression in tumor tissue argues for a reliable T cell marker to determine the effector functional state of cytotoxic lymphocytes (T cells or NK cells). We acknowledge that other immune cells may also express NKG7 and contribute to tumor immunity. To that end, our future studies would evaluate the anti-tumor effect of each subset of human immune cells transduced with NKG7 mRNA in treatment of human tumor xenografts in immunodeficient mice. Although several clinical trials based on the use of tumor infiltrating lymphocytes for solid cancer have been performed^60–62, the overall responses varied among patients with advanced solid cancer. Interestingly, some studies demonstrate that an inflammatory immune environment or an active interaction of immune cells at baseline is strongly associated with response to TIL therapy^63,64. To that end, NKG7 expression as our data show, could be a reliable predictive marker worth including in future clinical trials utilizing TIL-based immunotherapy for solid cancers. Additionally, inclusion of NKG7 expression in CAR-T cell therapies may also increase their effectiveness in tumor elimination. To better understand how Nkg7 specifically affects human CD8⁺ T cell responses to cancer, the future use of tools such as humanized mouse models may prove valuable.

Dynamic coordination of lysosomal homeostasis and metabolic signaling is necessary for CD8⁺ T cells not only to adapt to harsh environmental cues but also to develop lytic granule-mediated cytotoxic responses. Here, we identify NKG7 as a CD8⁺ T cell-specific lysosomal gatekeeper regulating lysosomal homeostasis and metabolism. Physiologically, the NKG7-mediated mTORC1 regulation is required for durable and long-lasting CD8⁺ T cell responses in the contexts of viral infection and tumor infiltration. Overall, these data open the possibility that NKG7 expression in CD8⁺ T cells can be explored to strengthen various immunotherapeutic approaches.

Methods

We confirm that all the mouse and human studies performed in this work comply with the relevant ethical regulation of the Mayo Clinic Institutional Animal Care and Use Committee and Institutional Review Board, respectively.

Cells, Media, and Cell Culture

Reagents for cell culture were purchased from Corning unless stated otherwise. DMEM or RPMI 1640 (ΜP Biomedicals or ThermoFisher) media were supplemented with final concentrations of 10% FBS (ThermoFisher), 2 mM L-glutamine, 100 IU/mL penicillin, and 100 μg/mL streptomycin to make complete medium. Jurkat human CD4⁺ T cells (ATCC) were cultured in complete RPMI medium. Human embryonic kidney 293 T cells (293 T; ATCC) were cultured in complete DMEM medium. Transduced 293 T cells were cultured in complete DMEM supplemented with 5 μg/mL Blasticidin S HCl. YTS human NK cells were cultured in YTS medium (complete RPMI medium additionally supplemented with final concentrations of 1x MEM non-essential amino acid, 1 mM sodium pyruvate, and 10 mM HEPES). Human peripheral blood mononuclear cells (PBMC) were isolated from whole blood by Ficoll-Paque PLUS density gradient centrifugation (Cytiva). Human CD8⁺ T cells or NK cells were isolated from PBMCs by negative selection according to the manufacturer’s protocols (Stemcell), where purities of isolated cells were greater than 95%. Purified cells were either immediately used for Cas9/RNP transfections or expanded as follows. CD8⁺ T cells were activated by plating cells into a plate pre-coated with 1 μg/mL anti-CD3 (clone OKT3; BioLegend) and anti-CD28 (clone CD28.2; BioLegend) and expanded in complete RPMI medium supplemented with 100 U/mL recombinant human IL-2 (PeproTech). T cells were used for functional analysis between 2 and 3 weeks after initial activation and expansion in IL-2. NK cells were maintained in complete RPMI medium supplemented with 40 U/mL recombinant human IL-2 and 20 ng/mL recombinant human IL-15 (PeproTech). MC38 (Colon adenocarcinoma) was from MIlliporeSigma (SCC172) and has been maintained as previously described⁶⁵.

Mice

Male and female 6- to 8-week-old C57BL/6 (B6) and B6.SJL (expressing the CD45.1 allele) mice were purchased from Jackson Laboratory (Stock #000664 and #002014, respectively) and allowed to acclimate to our housing facilities for at least one week. Nkg7^fl/fl mice were generated by inGenious Targeting Laboratory (Ronkonkoma, NY). A 10.5 kb region used to construct the targeting vector was first subcloned from a positively identified C57BL/6 BAC clone (RP23-339C21) using homologous recombination. The region is designed in such a way that the long homology arm (LA) extends 6.2 kb upstream of the distal LoxP cassette and the short homology arm (SA) extends 2.4 kb 3’ to the Neo cassette. The distal LoxP is positioned 323 bp upstream of exon 1. The LoxP-FRT flanked Neo cassette is inserted 373 bp downstream of exon 4. The targeted region is 1.9 kb including exons 1, 2, 3 and 4. PCR screening for the deletion is accomplished using the following oligonucleotides: PNDEL1 5’-GTGGCACAGGATCTCTCAGG-3’ and PNDEL2 5’-GATCAAGGGCTTGCCTTAGC-3’. LCMV H2-D^b gp33-specific TCR transgenic P14 mice were obtained from Dr. Stephen C. Jameson (available at Taconic – model #14983). P14 mice were fully backcrossed to B6 mice, with introduction of CD45.1 and CD45.2 congenic markers for identification. CD8⁺ T cell-specific NKG7 transgenic (referred to as “CD8-NKG7 Tg mouse” in this study) mice were generated by crossing E8I CD8-Cre mice (C57BL/6-Tg (Cd8a-cre)1Itan/J, Strain stock #: 008766; purchased from the Jackson Laboratory) with CAG2-NKG7-Rosa26-IRES-tdTomato mice (referred to as “Rosa26-NKG7 mouse” in this study; generated by inGenious Targeting Laboratory [Ronkonkoma, NY]). A LoxP flanked Neo stop selection marker was cloned downstream of the pCAG promoter sequence and followed by a NGK7-IRES-tdTomato-WPRE-BGHpA cassette. The targeting vector contains a short homology arm (SA) with 1.1 kb ROSA26 genomic sequence and a 4.3 kb long homology arm (LA) downstream sequence to guide vector integration. The vector was constructed such that upon Cre-Lox excision of the Neo-Stop cassette, the expression of a NKG7-IRES-tdTomato cassette will be driven by the pCAG promoter in the ROSA26 locus. PCR screening for the NKG7 transgene is accomplished using the following oligonucleotides: WPRE SQ1 5’-TCATGCTATTGCTTCCCGTATGGC-3’ and ROSA SQ2 5’-TGCTTACATAGTCTAACTCGCGAC-3’. In these mice, CD8 promotor driven Cre enzyme will remove a stop codon within a new construct ahead of the NKG7 gene in CD8⁺ T cells to enable bicistronic expression of NKG7 and tdTomato. All mice were maintained under specific-pathogen-free conditions at Mayo Clinic Rochester and Arizona, were kept in filler-top cages with access to food pellets and water under controlled ambient temperatures (20-22 degrees C) and relative humidity (30-70%), and a 12 h light/12 h dark cycle. In all experiments, mice were randomly assigned to experimental groups. The mice were euthanized by carbon dioxide according to the approved protocols. All animal protocols were approved by the Institutional Animal Care and Use Committee at Mayo Clinic Rochester and Arizona (IACUC: A00006353-21 and A00005542-20, respectively).

LCMV Infection Studies

LCMV (Armstrong and Clone 13 strains) were maintained at -80 °C until infection and diluted to 2x10⁶ PFU/mL (Armstrong) and 2x10⁷ PFU/mL (Clone 13) in PBS. WT (sgCd19) and Nkg7-KO (sgNkg7) P14 cells were adoptively transferred (5x10⁴ cells each, intravenously) into naive wild-type mice, which were infected 6 hours later with LCMV-Armstrong (2x10⁵ PFU, intraperitoneally) or LCMV-Clone 13 (2x10⁶ PFU, intravenously). Sometimes, wild-type (Nkg7^fl/fl) or CD8-Cre Nkg7^fl/fl mice were infected with LCMV-Armstrong or LCMV-Clone 13 with similar viral loads. In some experiments, LCMV-infected mice were treated with Bafilomycin A1 (1 mg kg^-1 mouse, MilliporeSigma) between days 1-3 after infection as previously described⁶⁶. In other experiments, mice were treated with Rapamycin (75 µg kg^-1 mouse, MilliporeSigma) between days 1-7 after infection as previously described⁴³. At each collection time point, spleen was removed from each mouse, and single-cell suspensions were prepared using a cell strainer for flow cytometry analysis. For PFU measurements from LCMV-Clone 13 infected mice, kidneys were collected and processed, and PFUs were measured as described previously⁶⁷.

Plasmids and cloning

For transient expression, NKG7 cDNA encoding full-length human NKG7 (NCBI no. NM_005601.4) was cloned into the pCI2 mammalian expression plasmid (multiple cloning sites inserted into pCI plasmid [Promega]) with either FLAG-YFP or FLAG-YFP-TurboID sequence located in-frame at the N-terminus. cDNA encoding full-length human ATP6AP2 (NCBI no. NM_005765.3) was cloned into pCI2 plasmid, with the EE-mCherry sequence immediately following in-frame at the C-terminus. cDNA of ATP6V0d1 from pRK5-FLAG-ATP6V0D1 (Addgene, #87931)³³ was subcloned into pCI2 plasmid. For lentiviral transductions, NKG7 cDNA was cloned into a modified pLenti lentiviral transfer plasmid with FLAG-YFP sequence located in-frame at the N-terminus. The following plasmids were obtained from Addgene: pCMV delta R8.2 (#12263), pCMV VSV-G (#8454)⁶⁸, and pRK5-p18-HA (#42338)⁶⁹. All constructs were validated by Sanger dideoxy sequencing.

Preparation of CRISPR Cas9/RNP complexes

Reagents for CRISPR Cas9/RNP system were obtained from IDT unless stated otherwise. The Cas9/RNP complexes were prepared following the manufacturer’s instructions and as previously described^70,71. To prepare the Cas9/RNP complexes, chemically modified CRISPR RNA (crRNA) specific to the target gene (control: GTTCCGCGTTACATAACTTA; NKG7: TGGTTGAGACAAGCGGGCCG and GGGACCGGCAGAGCTCCATG; ATP6AP2: AGGAGAGCGGATCCCAGACG and TCCTACTCAGAGAATTGAGG) and trans-activating CRISPR RNA (tracrRNA) were resuspended in Nuclease-Free Duplex Buffer at a concentration of 100 μM. Equimolar amounts of crRNA and tracrRNA oligos were mixed and incubated at 95 °C for 5 minutes, followed by gradual cooling to room temperature for annealing. The resulting crRNA-tracrRNA duplexes are referred to as “gRNA” in this study. For the formation of Cas9/RNP complexes, 450 pmol of gRNA and 180 pmol of Cas9 protein (UC Berkeley/Macrolab) were mixed and incubated at room temperature for 10 minutes. Unless stated otherwise, two gRNAs targeting two different genomic regions of the same gene were mixed together (225 pmol/each) with 180 pmol of Cas9 protein to form the Cas9/RNP complex. The entire Cas9/RNP complex was used immediately for a single transfection. gNeg: negative control. For Cas9/RNP nucleofection into mouse P14 cells, single guide RNAs (sgRNA) for either Cd19 (sgCd19: GAAGAGACAGGUGAGGAGUC and UGUUGUGCUGCCAUGCCUCC), Nkg7 (sgNkg7: CCAGGAAGCUCACAGACACC and GACACAGAGCUUCUGUAUCC) or Atp6v0d1 (sgAtp6v0d1: CACCUACAGAGUACAGAUUA and GAGGUGACGCUUCAUUGGCC) (Synthego) and Cas9 protein (TrueCut^TM Cas9 v2, ThermoFisher) were similarly pre-complexed at room temperature for at least 10 minutes before transfection.

Transfections

Transfections of Cas9/RNP into primary human CD8⁺ T cells followed a previously described method⁷¹. In brief, 10 million CD8⁺ T cells were suspended in 20 μL of P2 nucleofection solution (Lonza) and mixed with the prepared Cas9/RNP complex (as described in the “Preparation of CRISPR Cas9/RNP Complexes” section above). Nucleofection was carried out using a 4D nucleofector system with EH-100 program. Transfections of Cas9/RNP into primary human NK cells were done similarly using CM-137 program. After nucleofection, cells were transferred into pre-warmed culture media and maintained. Experiments were performed 5-7 days after nucleofection. CRISPR Cas9/RNP nucleofection into naïve mouse P14 cells was performed as previously described⁷⁰. 1-10 million WT P14 cells were resuspended in 20 μL of P3 nucleofection solution (Lonza) and mixed with the prepared sgRNA/Cas9 complex (as described in the “Preparation of CRISPR Cas9/RNP Complexes” section above). The mixture was transferred to nucleofection cuvette strips and nucleofection was carried out using a 4D nucleofector system, using the DN-100 program. After nucleofection, prewarmed complete RPMI was used to transfer transfected P14 cells in 96-well plates. After 15 minutes of resting, P14 cells were adoptively transferred into recipient mice.

Transient transfection of DNA plasmids into Jurkat T cells was conducted using a BTX ECM 830 electroporator with a single pulse at 315 V for 10 ms. For each transfection, a total of 40 μg of plasmid was introduced into 8 million cells. Experiments were performed 2 days after transfection. Unless stated otherwise, 293 T cells were transfected using polyethylenimine (PEI, Polysciences) as follows. For transfection into 293 T cells to perform exogenous immunoprecipitation (IP) or immunofluorescence (IF) staining for microscopy, 293 T cells were plated in 6-well cell culture plates at a density of 0.5 million cells per well, 16 to 20 hours prior to transfection. For each well, a total of 3 μg of plasmid DNA was diluted in 125 μL of OptiMEM media (ThermoFisher), and 6 μg of PEI dissolved in water (stock concentration at 2 mg/mL) was added. After a 15-minute incubation at room temperature, the medium volume in each well was adjusted to 2 ml, and the mixture was added to the cells dropwise. The treated cells were then incubated at 37 °C for 150 minutes. Following incubation, the medium was replaced with pre-warmed fresh growth medium. When transduced 293 T cells were used for transfection, cells were plated in poly-L-lysine (PLL)-coated (0.1% in H₂O [w/v]; MilliporeSigma) 6-well cell culture plate at a density of 0.8 million cells per well, 16 to 20 hours before transfection. For each well, 2.5 μg total of plasmid DNA was transfected using Lipofectamine™ 2000 Transfection Reagent, according to manufacturer’s instructions (ThermoFisher). For TurboID-based proximity labeling assay, 293 T cells were plated in 10 cm culture dishes at a density of 4 million cells per dish, 16 to 20 hours prior to transfection. For each dish, a total of 20 μg FLAG-YFP-TurboID-NKG7 DNA plasmid (3 μg for FLAG-YFP-TurboID plasmid) was mixed with 64 μg of PEI in 800 μL of OptiMEM media. The final culture volume was adjusted with medium to 6 mL, and the mixture was added to cells as described above. Experiments were performed 1 day (immunofluorescence staining for microscopy) or 2 days after transfection.

For the production of VSV-G-pseudotyped lentivirus particles, 14 million 293 T cells were placed per 150 mm culture dish 16 to 20 hours before transfection. For each dish, 51 μg of transfer vector, 34 μg of pCMV delta R8.2 and 17 μg of pCMV VSV-G were mixed with 100 μg of PEI in 10 mL OptiMEM media. The final culture volume was adjusted with medium to 15 mL, and the mixture was added to cells as described above.

Immunofluorescence Microscopy

Reagents for immunofluorescence (IF) microscopy were purchased from ThermoFisher Scientific unless stated otherwise. Expanded human primary CD8⁺ T cells, human primary NK cells, or transfected Jurkat T cells were washed once and resuspended in serum-free RPMI. 165,000 primary CD8⁺ T or NK cells, or 65,000 Jurkat T cells were plated directly on PLL (poly-L-lysine)-coated (0.1% in H2O [w/v]; MilliporeSigma) 18 mm round coverslips (#1.5 thickness; Electron Microscopy Sciences) and incubated at 37 °C for 2 minutes. In the case of CD8⁺ T cells treated with vacuolin-1 or acetate Ringer’s solution, 120,000 cells were plated on the same type of coverslip and incubated at 37 °C for 15 minutes instead. Unless stated otherwise, the cells were then fixed with 4% paraformaldehyde (PFA; Electron Microscopy Sciences) in PBS for 18 minutes and permeabilized with 0.15% Triton-X 100 in PBS for 4 minutes at room temperature. For staining of VPS41, dynein intermediate chain (DIC), and KIF5B, samples were fixed in ice-cold methanol for 5 minutes at room temperature instead. In the case of transfected or transduced 293 T cells, cells were collected by trypsinization, and a total of 250,000 cells were plated in 1 well of a 12-well culture plate containing a PLL-coated 18 mm round coverslip (#1.5 thickness) 16 to 20 hours prior to the experiment. Upon treatment as indicated, the coverslips were fixed and permeabilized as described above.

Intracellular distributions of endogenous proteins were detected using the following antibodies: anti-LAMP1 (clone H4A3; Santa Cruz Biotechnology [SCBT]), anti-NKG7 (rabbit polyclonal was produced by Cocalico Biologicals Inc, Denver, PA)¹⁹, anti-α-Tubulin (clone DM1A; MilliporeSigma), anti-Rab7 (clone B-3; SCBT), anti-Rab7 (clone D95F2; Cell Signaling Technology [CST]), anti-Arl8b (Proteintech), anti-VPS41 (clone E-10; SCBT), anti-LAMP1 (clone D2D11; CST), anti-mTOR (clone 7C10; CST), anti-LAMTOR1 (clone D11H6; CST), anti-dynein intermediate chain (clone 74.1; MilliporeSigma), and anti-KIF5B (clone SUK-4; abcam). Signals were detected by incubation with goat-anti-rabbit IgG and goat-anti-mouse IgG conjugated with Alexa Fluor (AF) 488 or AF 568. The coverslips were mounted on glass slides using Antifade mounting reagent with DAPI. Confocal images were acquired on an Airyscan-equipped LSM800 confocal microscope with a 63x oil immersion objective of NA 1.4 (Zeiss). Unless stated otherwise, z-stack images were taken for all experiments with the z-interval determined based on Nyquist sampling. In the case of imaging 293 T cells or CD8⁺ T cells treated with vacuolin-1 or acetate Ringer’s solution, a single focal plane of image was determined and obtained to best visualize LAMP1 signal. The acquired raw Airyscan images were processed with Zen Blue software (version 2.6; Zeiss) and exported to TIFF format. All images within each experimental set were taken using fixed acquisition settings.

For IF staining in cystectomy tissue, FFPE tissues were sectioned at 5 microns and tissue sections were deparaffinized in Histoclear twice and rehydrated through decreasing gradient of ethyl alcohol. For antigen retrieval, sections were heated in a 10 mM citrate buffer (pH 6.0) for 20 minutes at 95 °C in a steamer. Cell permeabilization was performed by incubating tissue slides with 0.5% Triton X-100 for 10 minutes at room temperature. After TBST washes, nonspecific antibody binding was blocked by incubating slides with 1% BSA in PBS with 1:60 goat serum at room temperature for 30 minutes. Slides were then incubated overnight at 4 °C with a monoclonal rabbit anti-NKG7 antibody (clone 8H3/8K3; in-house produced [Fusion Antibodies]) and mouse anti-CD8 antibody (clone RIV11; 1:500). After TBST washes, sections were incubated with goat anti-rabbit conjugated with AF 488 (1:1000) and donkey anti-mouse conjugated with AF 647 (1:1000) at room temperature for 60 minutes. Sections were washed in TBST and mounted using Antifade mounting reagent with DAPI. Fluorescence images were taken with a Leica DMI8 microscope equipped with 20x and 40x objectives.

Proximity ligation assay

Proximity ligation assay (PLA) was performed according to the manufacturer’s protocols (MilliporeSigma) and as previously described³⁹. Samples were fixed and permeabilized as described in Immunofluorescence Microscopy section. Coverslips were incubated with ATP6V0A4 (Proteintech, 1:200 dilution) and ATP6V1B1 (clone OTI6D2, Origene, 1:500 dilution) overnight at 4 °C. Coverslips were washed twice with 1x PBS and incubated with PLA secondary antibodies (1:5 dilution, anti-rabbit MINUS + anti-mouse PLUS) for 60 minutes at 37 °C. After two washes with 1x PBS, coverslips were incubated for ligation reaction for 30 minutes at 37 °C. Samples were washed twice with 1x PBS, and incubated for FarRed amplification reaction for 100 minutes at 37°C. Upon two washes with 1x PBS, coverslips were incubated with Hoechst 33342 (4 μg/mL) and rhodamine phalloidin (1:400) at room temperature for 20 minutes. Coverslips were washed and mounted for confocal microscopy. Z-stack images based on Nyquist sampling were acquired on LSM800 confocal microscope with a 63x oil immersion objective of NA 1.4 (Zeiss).

Image analyses

Unless stated otherwise, image analyses were done using Imaris software (Oxford Instruments, version 9.5 or higher). To determine lysosome number and size of human primary CD8⁺ T cells, human primary NK cells, or Jurkat T cells, LAMP1 staining was first defined using a fixed intensity threshold for all images within each experimental set. Individual lysosome (spot) was defined by utilizing ‘Spots’ algorithm on LAMP1 staining channel (with estimated XY diameter set as 0.15 μm). Enlarged/aggregated lysosome (cluster) was defined using ‘Surfaces’ algorithm with surface grain size set as 0.0426 μm and volume greater than 0.4 μm³. Any individual spot located within the defined cluster was considered as part of the cluster and removed. Total number of spots and clusters were counted and the ratio of cluster/spot calculated. In the case of 293 T cells, lysosome number and size were quantified similarly as above with the following settings: ‘Spots’ (estimated XY diameter set as 0.4 μm) and ‘Surfaces’ (surface grain size: 0.0426 μm, area greater than 4.5 μm²). Spatial correlations among intracellular proteins within the 3D volume (or 2D area in the case of 293 T cells) was determined by calculating Pearson’s coefficient using the ‘Coloc’ function within Imaris software. To measure lysosome diameter of CD8⁺ T cells treated with vacuolin-1 or acetate Ringer’s solution, LAMP1 staining was first defined using a fixed intensity threshold for all images within each experimental set. Lysosome diameter was defined as the longest linear length of individual LAMP1⁺ organelle, and average lysosome diameter of a cell was determined after measuring diameters of 10-15 (or max number of) lysosomes within a cell. Average lysosome diameter of 10-15 cells within each experimental group was then determined to obtain final average lysosome diameter. Number of PLA spots within a cell was quantified and automated using a customized macro program within ImageJ software (version 1.45 s; National Institutes of Health). With maximum z-projection images, cell boundary was determined based on phalloidin stain, and the number of PLA puncta above the defined intensity threshold was counted. The determined threshold setup was applied to all images within each experimental set.

Cell treatment

Lysosome recovery of human CD8⁺ T cells upon treatment of either vacuolin-1 or acetate Ringer’s solution (80 mM NaCl, 70 mM C₂H₃NaO₂, 5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 2 mM NaH₂PO₄, 10 mM HEPES, 10 mM glucose and 0.5 mg/mL bovine serum albumin) was examined similarly as previously described²⁶. 3 million cells were collected and washed twice with complete RPMI medium. For vacuolin-1 treatment, cells were treated with 10 μM vacuolin-1 (MilliporeSigma) in complete RPMI and incubated at 37 °C for an hour. In the case of acetate Ringer’s solution, cells were incubated at 37 °C for 40 minutes. After incubation, cells were washed twice and rested in culture medium. At each time point, 300,000 cells were collected and prepared for IF microscopy samples. For nocodazole treatment to disrupt microtubule polymerization, human CD8⁺ T cells were incubated with 10 μM nocodazole (MilliporeSigma) in complete RPMI at 37 °C for an hour prior to processing for IF microscopy. For bafilomycin A1 (BafA1; CST) treatment to inhibit v-ATPase, cells were washed twice and incubated with 10 nM BafA1 in complete RPMI medium for 16 hours prior to processing for IF microscopy. For 2-NBDG labeling of CD8⁺ T cells, 1-2 million cells were collected and washed once with glucose-free complete RPMI medium (MP Biomedicals; supplemented with final concentrations of 10% dialyzed FBS, 2 mM L-glutamine, 100 IU/mL penicillin, and 100 μg/mL streptomycin). Cells were then resuspended with glucose-free complete RPMI medium with 50 μM 2-NBDG. After 30-minute incubation at 37 °C, cells were washed once with glucose-free complete RPMI medium and run for flow cytometry to measure 2-NBDG signal (using FITC channel). For proximity labeling of TurboID-expressing cells, 293 T cells transfected with TurboID DNA plasmids (FLAG-YFP-TurboID or FLAG-YFP-TurboID-NKG7) were washed once with complete DMEM medium. Cells were then replaced with complete DMEM containing 50 μM biotin (MilliporeSigma) and incubated a cell culture incubator (37 °C, 5% CO₂) for 10 minutes. Cells were washed twice with ice-cold PBS and processed.

Seahorse analysis

Reagents were purchased from Agilent unless stated otherwise. Seahorse Xfe96 Bioanalyser was used to determine oxygen consumption rate (OCR) and extracellular acidification rate (ECAR). CD8⁺ T cells were washed in XF Base media (Seahorse XF RPMI medium with 2 mM glutamine, 10 mM glucose, 1 mM sodium pyruvate, and 5 mM HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid), pH 7.4 at 37 °C) for OCR or XF RPMI media with 2 mM glutamine for ECAR before being plated onto Seahorse cell culture plates coated with Cell-Tak (Corning) at 1×10⁵ cells/50 uL per well. The cells were allowed to adhere to culture plate for 1 hour and the assay media was topped to 180 µL. The OCR was measured via Seahorse Mito Stress assay, with addition of oligomycin (2 µM), carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP; 1.5 µM) and Rotenone and Antimycin (1.0 μM) as the final concentration per well. The ECAR was measured with addition of 10 mM glucose, 2 µM oligomycin, and 50 nM 2-deoxy-D-glucose (2-DG) as the final concentration per well. Assay parameters were measured as follows: 3-minute mix, no wait, 3-minute measurement, repeated 3-4 times at basal and after each addition. SRC was calculated as OCR at maximum rate (OCR_Max) - OCR in basal state (OCR_Bas). Mitochondrial ATP production was calculated by subtracting the minimum respiration rate after oligomycin injection from the basal respiration rate before oligomycin injection.

Lysosomal pH measurement

Lysosomal pH was determined using ratiometric LysoSensor Yellow/Blue DND-160 dye (ThermoFisher) as previously described⁷². In the case of 293 T cells, 0.1 million cells were plated onto PLL-coated 12-well plate. Next day, cells were incubated with complete DMEM medium with or without 7 μM BafA1 for 16 hours. Cells were then resuspended with complete DMEM medium with 2 nM LysoSensor Yellow/Blue DND-160 for 5 minutes at 37 °C. After two washes with 1x PBS, fluorescence intensities of cells at shorter wavelength (440 nm) and at longer wavelength (540 nm) were measured using microplate reader. To generate pH calibration curve, LysoSensor-labelled cells (prepared in parallel with the experimental groups) were incubated with pH calibration curve buffer (pH 3.5–7.0) at 37 °C for 10 min. For cells treated with each buffer, fluorescence intensity ratio of emission intensity at shorter wavelength (440 nm) to emission intensity at longer wavelength (540 nm) was plotted to generate pH calibration curve. Based on this curve, pH of each group of cells were determined based on the calculated fluorescence intensity ratio. Lysosomal pH of YTS cells was determined in a similar process as described above using 10 μM BafA1 and 2.5 nM LysoSensor Yellow/Blue DND-160 instead. To measure fluorescence intensity of each group of cells, 0.11 million processed cells were plated onto black-walled and clear-bottom 96-well microplate. Relative lysosomal acidity was measured and compared using LysoSensor Green DND-189 dye. For LysoSensor labeling of human and mouse CD8⁺ T cells and Jurkat cells, 1 million cells were washed once with complete RPMI medium. Cells were then resuspended with complete RPMI medium with 0.125 μM LysoSensor Green DND-189 (ThermoFisher). After 3-minute incubation at 37 °C, cells were spun down and run for flow cytometry to measure the LysoSensor Green signal (using FITC channel).

T cell activation

To examine T cell responses during T cell receptor (TCR)-induced CD8⁺ T cell activation and proliferation, 10 to 30 million purified human CD8⁺ T cells treated with Cas9/RNP were washed once and resuspended with complete RPMI medium supplemented with 100 U/mL recombinant human IL-2 (100 U/mL) at 3 million cells/mL. 6 to 9 million resuspended cells were then plated on a 6-well culture plate pre-coated with 1 μg/mL anti-CD3 (clone OKT3) and anti-CD28 (clone CD28.2) in the presence or absence of 50 nM rapamycin (Selleckchem.com). At each time point, 1 to 3 mL of cell culture was collected, washed once with PBS, and either lysed with lysis buffer-1 (see “Immunoblotting” section) for immunoblot, processed for flow cytometric intracellular staining (see “Flow Cytometric Analyses” section), or processed for 2-NBDG labelling (See “Cell Treatment” section). In the case of human T cell activation from healthy donors (Fig. 2e), human PBMCs isolated from healthy donors were activated with beads immobilized with anti-CD3 and CD28 antibodies (ThermoFisher). After 48 hours of incubation, beads were removed from cell culture using a magnet and cells were used for analysis as previously described¹⁹. For activation of mouse T cells via immunization (Fig. 2d), each mouse received a one-time injection (i.p.) of 0.5 mg of OVA protein (MilliporeSigma) and 50 µg of poly(I:C) (Novus Biologicals) in a total volume of 200 µL as previously described⁶⁵. The primed CD8⁺ T cells were isolated from the spleen on day 7 after immunization and used for flow cytometry analysis.

To examine phosphorylation of ribosomal S6 protein upon TCR stimulation (Supplementary Fig. 3e, f), 5 to 10 million purified human CD8⁺ T cells treated with Cas9/RNP were washed once and rested in serum-free RPMI at 37°C for one hour. Subsequently, the cells were resuspended in 200 μL of stimulation medium (serum-free RPMI containing 0.5% bovine serum albumin (BSA) [w/v]) for each stimulation condition. Anti-CD3 (clone OKT3) antibody was added to the resuspended cells with final concentration of 5 μg/mL, and the cells were incubated on ice for 10 minutes. Cells were then washed and kept on ice until stimulation. The cells were stimulated by resuspending them in 100 μL of stimulation medium containing 20 μg/mL of goat anti-mouse IgG antibodies (Jackson ImmunoResearch), to crosslink pre-bound anti-CD3 antibodies, and incubated at 37°C for various time-points, up to 30 minutes. To halt the stimulation, the cells were quickly washed once with ice-cold PBS and fixed with fixation buffer (BioLegend) for flow cytometric intracellular staining.

Immunoblotting

All reagents were purchased from ThermoFisher Scientific unless stated otherwise. Cells were rinsed with ice-cold PBS once and then treated with lysis buffer-1 (containing 25 mM HEPES [pH 7.9], 25 mM NaCl, 0.5% NP-40, 1 mM EDTA, 0.5 mM CaCl₂, 1 mM PMSF, 10 μg/mL leupeptin, and 5 μg/mL aprotinin) on ice for 20 minutes. The lysates were clarified by centrifugation at 13,000 x g for 10 minutes at 4 °C, and the protein concentration was determined using the Bradford assay (Bio-Rad). A total of 25 μg of protein was mixed with SDS-sample buffer and heated at 95 °C for 5 minutes. The prepared samples were separated on an SDS-PAGE gel and transferred onto PVDF membranes. The membranes were blocked with 4% BSA in TBS, and then incubated with the following primary antibodies (diluted in 2% BSA in TBS containing 0.1% Tween 20): anti-S6 (pS235/36; clone D57.2.2E; Cell Signaling Technology [CST]), anti-mTOR (clone 7C10; CST), anti-GAPDH (clone GT239; GeneTex), anti-4E-BP1 (pT37/46; clone 236B4; CST), anti-p70S6K (pT389; clone 108D2; CST), anti-p70S6K (clone 49D7; CST), anti-S6 (clone 5G10; CST), anti-FLAG (clone M2; MilliporeSigma), anti-EE (clone Glu-Glu; BioLegend), anti-ATP6V0D1 (clone EPR18320-38; abcam), anti-NKG7 (rabbit polyclonal), anti-Cathepsin D (R&D Systems), anti-ATP6V1A (Novus Biologicals), anti-ATP6V1E1 (Proteintech group), anti-ATP6AP2 (Novus Biologicals), anti-ATP6V0A1 (Novus Biologicals), anti-LAMTOR1 (clone D11H6; CST), anti-LAMP1 (clone H4A3; Santa Cruz Biotechnology [SCBT]), anti-α-Tubulin (clone DM1A; MilliporeSigma), anti-RagA (clone D8B5; CST), anti-LAMTOR3(clone D38G5; CST), anti-HA (clone C29F4; CST), and anti-Biotin (Jackson ImmunoResearch). Signals were detected by incubation with appropriate secondary antibodies: Peroxidase mouse anti-goat IgG (SCBT), goat anti-mouse IgG, goat anti-rabbit IgG, mouse anti-rabbit IgG (light chain specific), goat anti-rabbit IgG (Fc fragment specific), goat anti-mouse IgG (light chain specific), or goat anti-mouse IgG (Fcγ fragment specific) (all from Jackson Immunoresearch). Enhanced chemiluminescent (ECL) substrate was applied and chemiluminescent signals were detected using the ChemiDoc Touch Imaging System (Bio-Rad). Uncropped and unprocessed scans of all immunoblots presented in this study can be found in the Source Data file.

The densitometric analysis of immunoblots was conducted using ImageJ software (version 1.45 s; National Institutes of Health). In experiments involving a single time point, the band intensity of the protein of interest is normalized to the corresponding loading control, then the relative intensity was normalized to that of the control group (defined as a relative unit; RU). In experiments involving multiple time points, the band densitometry was initially normalized to the corresponding loading control at each time point. Subsequently, the change in the normalized signal compared to time point 0 within the control group was graphed using Prism version 9.5 software (GraphPad). The area under the curve (AUC) was calculated for each group over time and then normalized to the AUC of the control group, which was defined as the RU. For the densitometric quantification of LAMTOR1-HA co-IP (Fig. 5l), intensity of co-IP band was normalized based on the intensity of corresponding input band as well as the intensity of corresponding HA-IP band.

Flow cytometric analyses

All antibodies used for flow cytometric staining are purchased from BioLegend unless stated otherwise. For intracellular staining, samples were permeabilized and stained according to the ‘Intracellular Flow Cytometry Staining protocol’ from BioLegend unless stated otherwise.

To detect phosphorylation of S6 (p-S6) and p-AKT in TCR-activated human CD8⁺ T cells (Fig. 2c and Supplementary Fig. 3e, f, i), combinations of the following antibodies were used for staining: AF 647 anti-S6 (pS235/36; clone D57.2.2E; CST), AF 647 anti-S6 (pS240/44; clone D68F8; CST), PE anti-AKT (pT308; clone D25E6; CST), PerCP anti-human CD8 (clone SK1; BD), and AF 488 anti-human CD3 (clone SK7). Percent positive or mean fluorescence intensity (MFI) for p-S6/-AKT was measured for the gated population. In the case of examining p-S6 of human CD8⁺ T cells upon T cell activation among PBMCs (Fig. 2e), PBMCs were thawed, washed in complete RPMI, and then resuspended in FACS buffer (1x PBS, 2 mM EDTA, and 3% FBS) at a concentration of 1 x 10⁶ cells / 100 μL. Cells were then stained with anti-human CD8-PE-Cy7 (BD) and anti-human NKG7-AF488 (clone 8H3/8K3, conjugated with Zenon Rabbit IgG labeling Kits; ThermoFisher).

To examine p-S6 of mouse T cells activated via immunization (Fig. 2d), isolated CD8⁺ T cells from spleen were similarly stained as above with anti-mouse CD3 (clone 145-2C11), anti-mouse CD8-BV421 (clone 53-6.7), anti-CD44-BV570 (Clone IM7), anti-CD62L-PE/Cy7 (Clone:MEL-14), and anti-S6 (pS235/36; clone D57.2.2E; CST) in FACS buffer. Extracellular and intracellular staining on ex vivo mouse lymphocytes from spleens were performed as previously described⁶⁷. Tetramers to identify LCMV-specific CD8⁺ T cells (H-2D^b-gp33 tetramers conjugated with PE-Streptavidin) and OVA-specific CD8⁺ T cells (H-2K^b-SIINFEKL tetramers conjugated with PE-Streptavidin) were obtained from Yerkes NIH Tetramer Core. Among LCMV-specific CD8⁺ T cells, the following markers were used to distinguish respective populations: naïve T (CD44^-CD62L⁺), central memory T (T_CM: CD44⁺CD62L⁺), memory precursor (MP: CD127⁺KLRG1^-), terminal effector (TE: CD127^-KLRG1⁺), and stem-like exhausted (Ly108⁺Tim3^-). Adoptively transferred P14 cells were identified using a combination of CD45.1 and CD45.2 monoclonal antibodies. For p-S6 detection in mouse T cells, cells were stained with surface markers then fixed and permeabilized with the True-Nuclear™ Transcription Factor Buffer Set (BioLegend). Samples were then stained with anti-S6 (pS235/36; clone D57.2.2E; CST) antibody for 1 hour at room temperature. For intracellular staining of IFN-γ, Perforin, and Granzyme B, anti-IFN-γ (clone XMG1.2), anti-perforin (Clone S16009A), and anti-granzyme B- PE/Dazzle™ 594 (Clone: QA18A28) were used for intracellular staining. The degranulation was detected by staining with anti-CD107a-APC antibody (Clone 1D4B).

For most of flow cytometry experiments, Ghost Dye™ Red 780 (Cytek) or Zombie Fixable Viability Kit (BioLegend) were used to distinguish between live and dead cells. All samples were acquired on a BD FACSCanto II or FACS Symphony instruments with FACSDiva software, a CytoFLEX LX with CytExpert software (Beckman Coulter), or a Cytek Aurora with SpectroFlo software. Analyses were performed using the FlowJo software package versions 9.96 or higher (TreeStar). The gating strategies of flow cytometry data presented in this study can be found in Supplementary Fig. 8.

Lentiviral transduction

48 hours after transfection of lentiviral packaging plasmids (see “Transfections” above), the culture supernatants containing lentiviral particles were collected daily for 3 days and kept at 4 °C. The combined supernatants were then filtered using a 0.45 μm PVDF filter (MilliporeSigma) and concentrated by centrifugation at 18,000 x g for 3 hours at 4 °C. The resulting pellet was resuspended in OptiMEM media and stored at -80 °C until needed. The concentration of lentiviral particles (measured as infectious units) was determined by titrating the concentrated particles onto 293 T cells. An appropriate infectious unit for the specific target cells was selected to achieve transduction of approximately 5 to 15% of the cells. 293 T cells were placed with complete DMEM containing lentiviral particles and 8 μg/mL polybrene (MilliporeSigma) in a cell culture incubator (37 °C, 5% CO₂). After 16 to 24 hours of incubation, medium was replaced with fresh growth medium. To select and maintain 293 T cells stably expressing NKG7, blasticidin (ThermoFisher) was added at a final concentration of 5 μg/mL in the growth medium 7 days after transduction.

Amino acid starvation and stimulation

Reagents were obtained from Corning unless stated otherwise. Transduced 293 T cells were starved of amino acid and stimulated as previously described³⁰. Briefly, cells were washed twice and rested in amino acid-free complete RPMI medium (MyBioSource.com; supplemented with final concentrations of 10% dialyzed FBS [ThermoFisher], 2 g/L D-glucose [ThermoFisher], 100 IU/mL penicillin, and 100 μg/mL streptomycin) at 37 °C for 50 minutes. RPMI 1640 amino acid solution (50x stock; MilliporeSigma) and L-glutamine was applied to final concentrations of 1x and 2 mM, respectively, and incubated at 37 °C for various time-points, up to 20 minutes. At indicated time point, samples were either fixed for IF microscopy or lysed to examine phosphorylation of signaling proteins.

Immunoprecipitation and pull-down assays

For immunoprecipitations (IP), 2 μg of rabbit monoclonal anti-human NKG7 antibody (clone 8H3/8K3; in-house produced [Fusion Antibodies]) or mouse monoclonal anti-FLAG antibody (clone M2; MilliporeSigma) were bound to 30 μL of Protein G Agarose (ThermoFisher). Anti-HA affinity matrix (Roche) was used for HA-IP. For endogenous NKG7 IP, expanded CD8⁺ T cells were lysed in lysis buffer-1 and 0.75 to 1 mg clarified lysate was rotated with the antibody-bound agarose at 4 °C for 16 hours. To perform exogenous IP, transfected 293 T cells were lysed in lysis buffer-2 (consisting of 1 M HEPES [pH 7.2], 50 mM CH₃CO₂K, 1 mM EDTA, 200 mM D-sorbitol, 0.1% Triton X-100, 1 mM PMSF, and a protease inhibitor cocktail), and 0.3 to 0.5 mg of the clarified cell lysate was incubated at 4 °C for 2.5 to 4 hours in rotation with the antibody-bound beads. The protein complexes were subjected to 3 to 5 washes using the same lysis buffer used initially and resuspended in 30 μL of SDS-sample buffer. To confirm biotin labeling of the samples via TurboID (Supplementary Fig. 4b), biotin-treated cells were lysed in lysis buffer-1 and 0.5 mg clarified lysate was rotated with Neutravidin-conjugated agarose (ThremoFisher) at 4 °C for an hour. The protein complexes were washed 3 times using the same lysis buffer and resuspended in 50 μL of SDS-sample buffer. Unless stated otherwise, all samples were eluted after 5-minute incubation at 95 °C and resolved as described in the “Immunoblotting” section.

Mass-spectrometry Analysis of NKG7 Immunoprecipitates

Transduced 293 T cells stably expressing FLAG-YFP or FLAG-YFP-NKG7 were lysed in lysis buffer-1 and 1.8 mg clarified lysate was rotated with the 30 μL of pre-washed anti-FLAG magnetic beads (MilliporeSigma) at 4 °C for 16 hours. The samples were then washed 4 times with the same lysis buffer and 3 times with ice-cold PBS and then eluted twice with 50 µL of 5% acetic acid. After complete dried down on a speed vac, the eluates were resuspended in 20 µL of 20 mM Tris, pH 7.5/0.005% zwittergent 3-16 and subjected to reduction by 5 mM tris(2-carboxyethyl)phosphine (TCEP), alkylation by 10 mM iodoacetamide (IAA), and enzymatic digestion by trypsin/Lys-C overnight at 37 °C. The resulting tryptic digest was acidified and analyzed by LC-MS/MS on an Exploris 480 Orbitrap mass spectrometer (ThermoFisher). The acquired mass spectra were searched by SequestHT⁷³ against a UniProt human protein database (Ver. 2023-01) supplemented with 3 protein entries (A0A059PIR9 [YFP]; A0A059PIR9-FLAG [FLAG-YFP fusion]; Q16617-YFP [FLAG-YFP-NKG7 fusion]) using Proteome Discoverer (ThermoFisher, version 3.0). The search parameters included fixed modifications of carbamidomethylation on cysteine, variable modification of oxidization on methionine, and protein N-terminus acetylation. The intensities of proteins identified in both FLAG-YFP- and FLAG-YFP-NKG7-expressing 293 T cells were used for bioinformatics analysis of NKG7 interactome.

BioSITe analysis of NKG7-proximal proteins by TurboID

Biotin-treated cells (FLAG-YFP-TurboID or FLAG-YFP-TurboID-NKG7) were washed with ice-cold PBS three times before harvested in urea lysis buffer (9 M Urea, 50 mM TEABC, pH 8.5, Pierce Protease and Phosphatase Inhibitor tablet) for cell lysis by sonication. BioSITe analysis of NKG7-proximal proteins biotinylated by TurboID were carried out as previously described⁷⁴. Briefly, 10 mg of protein lysate sample was subjected to reduction by 5 mM DL-dithiothreitol, alkylation by 10 mM IAA, and enzymatic digestion by trypsin/Lys-C overnight at room temperature. The resulting tryptic peptides were acidified and desalted on a Sep-Pak C18 column prior to lyophilization. The purified peptides were resuspended in BioSITe capture buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.5% Triton X-100) and then incubated with 150 µg anti-biotin antibody (Bethyl Laboratories) conjugated to protein G beads on a rotator for 1 hour at 4 °C. The biotinylated peptides bound to the antibody on the beads were washed twice with BioSITe capture buffer, twice with 50 mM Tris (pH 7.5), and twice with ultrapure water. Samples were then eluted twice with 0.2% trifluoroacetic acid (TFA) and twice with 0.2%TFA/40% acetonitrile (ACN). The eluted biotinylated peptides were desalted using a C18 Empore stage tip and then analyzed by LC-MS/MS on an Exploris 480 Orbitrap mass spectrometer (ThermoFisher). The acquired mass spectra were searched by Andromeda⁷⁵ against a UniProt human protein database (Ver. 2021-03) on the MaxQuant proteomics analysis platform (Ver. 2.0.3.0)⁷⁶. The search parameters include fixed modifications of carbamidomethylation on cysteine, variable modification of oxidization on methionine, protein N-term acetylation, and biotinylation on lysine. The intensities of proteins and biotinylated peptides identified in FLAG-YFP-TurboID- and FLAG-YFP-TurboID-NKG7-transfected 293 T cells were used for bioinformatics analysis of NKG7 interactome.

Bioinformatic analysis of NKG7-interactome

TurboID and FLAG-NKG7-IP protein abundance data was used for downstream analysis with Rstudio using ggplot2, VennDiagram, RColorBrewer and ComplexHeatmap R packages. For the TurboID experiments, we selected candidate proteins that showed a Control (TurboID without NKG7) Intensity <2e6 to eliminate non-specific contributions and TurboID Intensity >3e6 to select significant biotinylated candidates. For the FLAG-YFP-NKG7-IP experiments, we calculated the ratio between Control (FLAG-YFP) and FLAG-YFP-NKG7 samples and selected for proteins showing a fold change > 10. Because NKG7 is mostly expressed in CD8⁺ T and NK cells we next selected for proteins that are expressed in CD8⁺ T cell subsets using the proteomic dataset from Aalderen et al.³². Finally, we determined the overlap between FLAG-NKG7 and TurboID experiments to identify the most promising interacting candidates. Additionally, we used the public NKG7 Y2H dataset and NKG7 available interactomes (BioGRID and String) to search for common proteins.

Lysosome and membrane fractionations

Lysosome fractionation of 293 T cells was performed as previously described⁷⁷. Briefly, 20 million 293 T cells were resuspended in 1 mL of 0.25 M sucrose solution with 10 mM Tris (pH 7.4) and homogenized. The homogenate was then centrifuged at 600 × g for 5 minutes at 4 °C to obtain the postnuclear lysate (PNL) supernatant. The PNL supernatant was further centrifuged at 12,000 × g for 10 minutes at 4 °C. The resulting supernatant was collected, and CaCl₂ was added to achieve a final concentration of 8 mM, followed by thorough mixing. The mixture was then centrifuged at 25,000 x g for 15 minutes at 4 °C, and the supernatant was collected as the non-lysosomal fraction (NLF). The pellets obtained were washed once with 150 mM KCl in 10 mM Tris (pH 7.4) and then resuspended in 150 μL of 150 mM KCl in 10 mM Tris (pH 7.4) to obtain the crude lysosomal fraction (CLF). Membrane fractionation of 293 T cells and expanded human CD8⁺ T cells were isolated using Mem-PER™ Plus Membrane Protein Extraction Kit (ThermoFisher).

Tumor growth and T cell analysis

C57BL/6 mice (control Rosa26-NKG7 with no CD8-Cre and CD8-NKG7 Tg mice) were subcutaneously injected with MC38 tumor cells (5 x 10⁵ cells/mouse). Tumor growth was evaluated every 2–3 days until ethical endpoints (tumor size = 1 mm³), when all mice were euthanized in compliance with animal care guidelines. Perpendicular tumor diameters were measured using a digital caliper (Carbon Fiber Composite, Fisher Scientific), and tumor sizes were calculated as length x width. On day 16 after tumor injection, tumor tissue and spleen were collected, and lymphocytes were isolated from each. To examine IFN-γ secretion by CD8⁺ T cells in spleen and tumor-infiltrating lymphocytes (TILs), both spleen cells and TILs were first treated with PMA (50 ng/mL) and ionomycin (500 ng/mL) (both from MilliporeSigma) in a cell culture incubator (37 °C, 5% CO₂) for 5 hours. After incubation, cells were stained of intracellular IFN-γ along with anti-CD3 and anti-CD8 and analyzed by flow cytometry, in which CD3⁺CD8⁺ T cells were gated for IFN-γ detection.

Immunohistochemistry of NKG7 in bladder cancer tissue

With the approval of the Mayo Clinic Institutional Review Board (IRB# 21-000078), we identified 158 patients who underwent radical cystectomy between 1990 and 2016. All patients did not receive neoadjuvant or adjuvant therapy. Our database was used to collect baseline clinicopathological variables such as age, gender, metastasis and recurrence status, tumor stage, histological type, lymph node (LN) status, and clinical follow-up data.

Tissue sectioning and IHC staining was performed at the Pathology Research Core (Mayo Clinic, Rochester, MN) using the Leica Bond RX stainer (Leica). FFPE tissues were sectioned at 5 microns and IHC staining was performed on-line. Slides were retrieved for 20 minutes using Epitope Retrieval 2 (EDTA; Leica) and incubated in Protein Block (Dako) for 5 minutes. Cell permeabilization was performed by incubating tissue slides with 0.5% Triton X-100 for 10 minutes. The rabbit monoclonal anti-NKG7 antibody (clone 8H3/8K3; in-house produced [Fusion Antibodies]) was diluted to 1:1000 in Background Reducing Diluent (Dako) and incubated for 15 minutes. The detection system used was Polymer Refine Detection System (Leica). This system includes the hydrogen peroxidase block, post primary and polymer reagent, DAB, and Hematoxylin. Immunostaining visualization was achieved by incubating slides for 10 minutes in DAB and DAB buffer (1:19 mixture) from the Bond Polymer Refine Detection System. To this point, slides were rinsed between steps with 1x Bond wash buffer (Leica). Slides were counterstained for 5 minutes using Schmidt hematoxylin and molecular biology grade water (1:1 mixture), followed by several rinses in 1x Bond wash buffer and distilled water. Once the immunochemistry process was completed, slides were removed from the stainer and rinsed in tap water for five minutes. Slides were dehydrated in increasing concentrations of ethyl alcohol and cleared in 3 changes of xylene prior to permanent cover-slipping in xylene-based medium.

T-cell infiltration and NKG7 staining was assessed by a pathologist blinded to clinical data. A 4-point scale was utilized: 0 (negative NKG7 expression), 1 (low expression), 2 (medium expression), 3 (high expression). The relationship between NKG7 expression and overall survival was evaluated using univariate and multivariate Cox proportional hazards regression models and presented using hazard ratios (HR) and 95% confidence intervals (CI). The Kaplan-Meier method was used to calculate overall survival probabilities following cystectomy, and differences in survival were assessed using the log-rank test. All tests were two-sided, and a P-value of <0.05 was considered statistically significant. Statistical analyses were done using GraphPad Prism version 10.

Pan-cancer association of NKG7 gene expression with patient overall survival

Using the online survival analysis tool KM-plotter⁷⁸, Kaplan-Meier estimates were used to estimate survival curves based on NKG7 expression in TCGA cancers: UCEC (N = 543 patients), READ (N = 165), CESC (N = 304), ESCA (N = 81), SARC (N = 259), STAD (N = 375), LIHC (N = 371), BLCA (N = 405), BRCA (N = 1090), HNSC (N = 500), OV (N = 374), LUAD (N = 513), LUSC (N = 501), PAAD (N = 177), KIRC (N = 530), KIRP (N = 288). The HR of each biomarker was calculated with univariate Cox proportional hazard model. NKG7 expression was converted to categorical variables and used to classify patients as high and low. The optimal cut-point value for each cancer was defined as the point with the most significant (log-rank test) split⁷⁹. HRs with 95% confidence intervals were calculated. Prism version 9.0.1 (GraphPad Software) was used for data visualization.

Statistical analysis

All statistical analyses were performed using GraphPad Prism version 8 or higher software, and statistical details of experiments can be found in the figure legends. All quantified results from fluorescent microscopic images are presented as SuperPlots⁸⁰.

In the case of LCMV infection studies, data were first subjected to the Kolmogorov-Smirnov test to assess normality of samples.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conceptualization, H.H, H.BdS., H.D., and D.D.B.; Methodology, H.H., J.Z., H.BdS., and H.D.; Formal Analysis, H.H., C.C., F.L., H.BdS., and H.D.; Investigation, H.H., J.B.H., D.T.B., Z.M., J.K.G., B.G.M., M.F.R.-Q., D.F.S., J.Z., K.E.M., A.B., D.S.A., A.L., R.G., R.P., H.BdS.; Resources, H.H., J.B.H., F.L., H.BdS.; Writing – Original Draft, H.H. F.L., H.BdS., H.D., and D.D.B., Writing – Review & Editing, H.H., A.P, H.L., F.L., H.BdS., H.D., and D.D.B.; Visualization, H.H., J.B.H., F.L., H.BdS., H.D., and D.D.B.; Supervision, H.H., A.P., H.L., F.L., H.BdS., H.D., and D.D.B.; Funding Acquisition, F.L., H.BdS., H.D. and D.D.B.

Peer review

Peer review information

Nature Communications thanks Bo Huang, Ken Inoki and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

Source data are provided with this paper and can be accessed at 10.6084/m9.figshare.28296227.v1. Proteomics datasets have been deposited in MassIVE and can be accessed at https://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=768764fa2bff4a3cbfcc403fe3fd1b44.

Competing interests

H.BdS. is an advisor for the International Genomics Consortium. The remaining authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-56931-6.