Vitamin B6 Competition in the Tumor Microenvironment Hampers Antitumor Functions of NK Cells

Abstract

Nutritional factors play crucial roles in immune responses. The tumor-caused nutritional deficiencies are known to affect antitumor immunity. Here, we demonstrate that pancreatic ductal adenocarcinoma (PDAC) cells can suppress NK-cell cytotoxicity by restricting the accessibility of vitamin B6 (VB6). PDAC cells actively consume VB6 to support one-carbon metabolism, and thus tumor cell growth, causing VB6 deprivation in the tumor microenvironment. In comparison, NK cells require VB6 for intracellular glycogen breakdown, which serves as a critical energy source for NK-cell activation. VB6 supplementation in combination with one-carbon metabolism blockage effectively diminishes tumor burden in vivo. Our results expand the understanding of the critical role of micronutrients in regulating cancer progression and antitumor immunity, and open new avenues for developing novel therapeutic strategies against PDAC.

Significance: The nutrient competition among the different tumor microenvironment components drives tumor growth, immune tolerance, and therapeutic resistance. PDAC cells demand a high amount of VB6, thus competitively causing NK-cell dysfunction. Supplying VB6 with blocking VB6-dependent one-carbon metabolism amplifies the NK-cell antitumor immunity and inhibits tumor growth in PDAC models. This article is featured in Selected Articles from This Issue, p. 5.

Figure 1.

PDAC cells create a vitamin B6-defective microenvironment that inhibits NK-cell activation. A, PCA plot of RNA-seq results of NK cells cocultured with HPNE, T3M4, CFPAC1, and Capan2. B, GSEA of inflammatory response genes based on RNA-seq data from NK cells were cocultured with PDAC cells, HPNE, or K562. C, Heat map of the mRNA expression of genes related to NK-cell cytotoxicity. D and E, Flow cytometry analysis showing the expression of NKG2A and TIGIT in NK (CD3⁻, NK1.1⁺) cells from KPC1245 orthotopic tumors or healthy mice blood. F, Dead cell percentage of CFPAC1 after coculture with NK cells under different conditions. CM, cells were cocultured in a CFPAC1-conditioned medium. FM, cells were cocultured in fresh medium. G and H, Expression of IFNγ and CD107a in NK cells from different conditions in F. I, Dying cell percentage of K562 upon coculturing with NK cells from different conditions. FM-NK, cells were cocultured in a fresh medium. CM-NK, cells were cocultured in the CFPAC1 CM. CM >3 kDa-NK, cells were cocultured in basal NK-cell medium with >3 kDa macromolecular components from CFPAC1 CM. CM <3 kDa-NK, cells were cocultured in <3 kDa molecular components from CFPAC1 CM. J and K, The percentage of IFNγ- and CD107a-positive cells in NK cells form different conditions as in I after coculturing with K562 cells. L, Partial least squares discriminant analysis (PLS-DA) plot of metabolites in NK-cell after coculturing with different cells. M, Top affected metabolic pathways in NK cells when cocultured with PDAC cells [co-PDAC (co-Panc1, co-T3M4, and co-Capan2) vs. co-CTL (NK only and co-HPNE)]. N and O, Relative intracellular levels of pyridoxine and pyridoxal phosphate (PLP) in NK cells after coculturing with the indicated cells. P and Q, Pyridoxine and PLP levels in plasma and tumor interstitial fluid from healthy tumor-free mice and mice with KPC1245 tumors. R, Serum PLP level from patients with pancreatic cancer or gallbladder stone. S, Dead cell percentage of K562 and HPNE cells after coculturing with NK cells that were precultured in media with different VB6 (pyridoxine) levels. Data, mean ± SEM. Unpaired Student t test (two-tailed) was used for D, E, and R. One-way ANOVA with Tukey multiple comparisons test was used for F–K, N–Q, and S. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

Figure 2.

Targeting VB6 and Kynurenine pathways enhances NK-cell antitumor immunity. A and B, Dead cell percentage of CFPAC1 or T3M4 cells upon coculture with NK cells and treated with or without IDO1 inhibitor (IDO1i), VB6 (PLP), or combination. C, The percentage of IFNγ- and CD107a-positive cells in NK cells upon coculturing with CFPAC1 and in the presence of control, IDO1i, VB6, or their combination. C, Dead cell percentage of T3M4 upon coculturing with NK cells and treating with or without IDO1i and VB6. D, Representative ultrasound images showing the largest cross-sections of the KPC1245 tumors upon treating with saline, IL15, VB6, IDO1i, VB6 + IDO1i, or VB6 + IDO1i + IL15 as described in Supplementary Fig. S2G. Scale bar, 10 mm. E, Quantification of KPC1245 tumor weight upon treating with saline, IL15, VB6, IDO1i, or their combinations. F, The ratio of tumor-infiltrating NK-cell (CD3⁻, NK1.1⁺) in total lived cells from KPC1245 tumors treated with IL15, VB6, IDO1i, and their combinations. G, The percentage of IFNγ, NKG2D, NKG2A, or TIGIT-positive cells in tumor-infiltrating NK cells in KPC1245 tumors with indicated treatment. H, Kaplan–Meier plots with the Mantel–Cox log-rank test for overall survival of mice orthotopically implanted with KPC1245 cells and treated with saline, IL15, VB6, IDO1i, or their combinations. I, Kaplan–Meier plots with the Mantel–Cox log-rank test for overall survival of mice implanted with KPC1245 cells and treated with saline or IDO1i and VB6, and with or without anti-NK1.1 antibody. Data, mean ± SEM. Two-way ANOVA with Tukey test was used for A–C. One-way ANOVA with Tukey multiple comparisons test was used for E–G. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. CTL, control.

Figure 3.

Vitamin B6 regulates glycogen metabolism. A and B, Relative intracellular levels of glucose 1-phosphate (Glc-1-P) and UDP-glucose in NK cells after being stimulated by HPNE, T3M4, CFPAC1, or Capan2. C, Representative transmission electron microscopy images of NK cells cultured with 25 mmol/L glucose without or with glycogen phosphatase inhibitor (GPI, 1,4-dideoxy-1,4-imino-D-Arabinitol, 400 μmol/L) treatment. GB, glycogen body. Scale bar: left two images, 500 nm; right image, 200 nm. D, Relative glycogen content in human NK cells (hNK), mouse NK cells, and the NK92 cell line after being stimulated by HPNE or mouse primary pancreatic epithelial cells (mPEC) for four hours, respectively. E, Dying cell percentage of HPNE after co-culturing with or without NK cells and GPI treatment for 20 hr (E:T, 2:1). F, The expression of IFNγ in NK cells after being cocultured with HPNE and treated without or with GPI. G, Relative glycogen content in NK cells after being stimulated with HPNE, T3M4, CFPAC1, or Capan2. The experimental design is described in Supplementary Fig. S3N. H and I, Relative intracellular levels of Glc-1-P and UDP-Glucose in NK cells after being stimulated by CFPAC1 without or with VB6 supplementation. J, Relative glycogen content in NK cells after being stimulated with HPNE, T3M4, or CFPAC1 cells and treated with VB6 or GPI. K, Dying cell percentage of HPNE, CFPAC1, and T3M4 cells upon coculturing with or without NK cells and with VB6 (PLP) or GPI treatment. L and M, The relative mRNA levels of glycogen phosphorylase (PYG) isoenzymes and glycogen synthesis (GYS) isoenzymes in NK cells from indicated culture conditions. N, Representative immunoblots showing the expression of GYS1 and PYGB in NK cells from indicated culture conditions. O, Representative immunoblot images showing the expression of GYS1 and PYGB in NK cells transfected with siRNA against GYS1 or PYGB. P, Dying cell percentage of K562 after coculturing with control, GYS1-knockdown, or PYGB-knockdown NK cells. Q and R, The expression of IFNγ and CD107a in control, GYS1-knockdown, or PYGB-knockdown NK cells after coculturing with K562. S, Dying cell percentage of CFPAC1 after coculturing with control, GYS1-, or PYGB-knockdown NK cells without or with VB6 treatment. Data, mean ± SEM. Unpaired Student t test (two-tailed) was used for D. One-way ANOVA with Tukey multiple comparisons test was used for A, B, E–I, P, and S. Two-way ANOVA with Tukey test was used for J, K, Q, and R. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. CTL, control.

Figure 4.

Glycogen-derived glucose supports glycolysis in NK cells with acute stimulation. A, Flow chart showing the experimental design for investigating the effects of VB6 and HPNE on glycogen metabolism alterations in NK cells. B, Relative pyridoxine, PLP, glucose-1-phosphate (Glc-1-p), and UDP-glucose (UDP-Glc) levels in NK cells from indicated treatment groups described in A. C, Relative levels of glycolytic intermediates in NK cells from indicated treatment groups described in A. D, The flow chart showing the experimental design for investigating the glucose and glycogen tracing in NK cells. E–G, Peak intensities of ¹²C-fraction of glucose-1-P and metabolic intermediates of glycolytic pathways in NK cells stimulated with or without HPNE for indicated time as described in D. G-6-P, glucose-6-phosphate; F-6-P, fructose-6-phosphate; 3PG, 3-phosphoglycerate; 2PG, 2-phosphoglycerate; PEP: phosphoenolpyruvate. H, Flow chart showing the experimental design for investigating the effects of glucose and GPI on NK-cell cytotoxicity. I, Dead cell percentage of K562 or HPNE cells upon coculturing with NK cells under different glucose concentrations. J, Dead cell percentage of K562 or HPNE cells upon coculturing with NK cells under normal or glucose-free conditions with or without GPI treatment. Data, mean ± SEM. One-way ANOVA with Tukey multiple comparisons test was used for B, C, I, and J. Two-way ANOVA with Tukey test was used for E–G. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

Figure 5.

PDAC cells maintain a high VB6 metabolic rate to promote cell growth. A, Relative pyridoxine levels in fresh alpha-MEM and conditioned media from NK92, K562, HPNE, and PDAC cells. B, Peak areas of pyridoxine and PLP in human NK cells (hNK), NK92, K562, HPNE, and PDAC cells. C, Schematic diagram showing products of metabolic labeling with ¹³C-labeled VB6 (pyridoxine). D and E, The fraction of ¹³C-labeled pyridoxine and PLP in total pyrido­xine and PLP pools in hNK, NK92, K562, HPNE, and PDAC cells cultured with ¹³C-labeled pyridoxine for 1, 6, 24, or 48 hours. F, The relative cell numbers of CFPAC1 and T3M4 cells upon culture with different VB6 concentrations for 72 hours on 2D assays. G, Quantification data of Cell-titer Glo cell viability assays for T3M4, PaTu, and CFPAC1 upon culturing with different concentrations of VB6 for 4 days in the 3D assay. H and I, Representative images and quantification data showing the growth of pancreatic cancer organoids PanC137 and Panc193 cultured with different concentrations of VB6 for 6 days. Scale bar, 1 mm (H). J and K, The relative level of pyridoxine and PLP in sera from KPC1245 tumor-bearing mice fed with standard (7 mg/kg VB6) or VB6-deficient diets. L, The weight of KPC1245 tumors from the mice fed with standard or VB6-deficient diets. Data, mean ± SEM. One-way ANOVA with Tukey multiple comparisons test was used for A, B, F, G, and I. Unpaired Student t test (two-tailed) was used for J–L. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

Figure 6.

Vitamin B6 regulates one-carbon metabolism in PDAC cells. A, Relative intracellular levels of PLP, glycine, cystathionine, glutathione (GSH), and SAM (S-adenosyl methionine) in CFPAC1 cells cultured with 0.1 mg/L or standard (1.0 mg/L) VB6 (pyridoxine). B and C, The relative cell numbers of CFPAC1 and KPC1245 cells cultured without (−) or with (+) VB6 (1.0 mg/L), glycine (100 mg/L), or formate (1 mmol/L). D, Represented immunoblots showing the expression of SHMT1 and SHMT2 in CFPAC1 cells without (siCTL) or with SHMT1 or SHMT2 knockdown [siSHMT1. siSHMT2, or siSHMT1&2 (double knockdown)] and VB6 supplementation. E, Relative cell number of CFPAC1 without or with SHMT1 and SHMT2 knockdown and VB6 supplementation. F, Representative immunoblots showing the expression of SHMT1 and SHMT2 in T3M4 cells with or without SHMT1, SHMT2 knockdown and VB6 supplementation. G, Relative cell numbers of T3M4 with or without SHMT1 or SHMT2 knockdown and VB6 supplementation. H, Relative cell numbers of CFPAC1 and T3M4 cells treated with SHIN1 (SHMT inhibitor). I and J, Representative images and quantification data showing the growth of PDAC organoids without or with SHMT inhibitor SHIN1 (10 μmol/L) treatment. Scale bar, 1 mm (I). K, Relative cell numbers of KPC1245 without (shCTL2) or with Shmt1&2 knockdown (shShmt1&2) and VB6 supplementation. L and M, Representative images and quantification data showing the growth of 3D spheroids of KPC1245 cells with or without Shmt1&2 knockdown and VB6 supplementation. N, Representative images of tumors derived from KPC1245 scrambled control or Shmt1&2 knockdown cells. O, The weight tumors derived from KPC1245 control (shCTL) or Shmt1&2-knockdown (shSHMT1&2) cells. Data, mean ± SEM. Unpaired Student t test (two-tailed) was used for A and J. One-way ANOVA with Tukey multiple comparisons test was used for B, C, E, G, H, and O. Two-way ANOVA with Tukey test was used for K and M. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

Figure 7.

VB6 supplementation combined withSHMT silencing impedes PDAC tumor growth. A, Representative immunoblots showing the expression of SHMT1 and SHMT2 in human primary NK cells cultured with different concentrations of VB6. B, Relative NK-cell viability upon treatment with SHIN1 (SHMT inhibitor). C, Representative immunoblots showing the expression of SHMT1 and SHMT2 in scrambled control, SHMT1-, SHMT2-knockdown human NK cells. D, Relative cell numbers of control, SHMT1-, and SHMT2-knockdown NK cells. E, Dead cell percentage of HPNE and K562 cells upon coculturing with control or SHMT1&2-knockdown NK cells. F, The percentage of IFNγ-positive cell in control, SHMT1-. or SHMT2-knockdown NK cells stimulated by HPNE or K562 cells. G and H, Relative live-cell number of CFPAC1 and KPC1245 cells upon coculturing with or without NK cells combined with indicated treatments. I, The weight of tumors derived from control (shCTL) or Shmt1&2-knockdown (shShmt1&2) KPC1245 cells in mice fed with VB6-free (No VB6), standard (7 mg/kg VB6), or VB6 high (70 mg/kg) diets and treated with or without IL15 and IDO1i. J, The percentage of tumor-infiltrating NK cells in total lived cells from tumors derived from control (shCTL) or Shmt1&2-knockdown KPC1245 cells and treated with or without IL15 and IDO1i. K and L, The percentage of granzyme B– and IFNγ-positive NK cells in total tumor-infiltrating NK cells in the tumors described in I. M, Kaplan–Meier plots with the Mantel–Cox log-rank test indicating survival of mice orthotopically implanted with control or Shmt1&2-knockdown KPC1245 cells and fed with standard (7 mg/kg) or high VB6 (70 mg/kg) diets with or without IL15 and IDO1 inhibitor treatments. N, Schematic summary graph of the main conclusion of the study. PDAC cells consume a large amount of VB6 to support their growth and reduce VB6 accessibility for other cells, including NK cells. VB6 deficiency leads to impeded glycogen breakdown and impaired effector functions in NK cells. Data, mean ± SEM. One-way ANOVA with Tukey multiple comparisons test was used for B and D–F. Two-way ANOVA with Tukey test was used for G–L. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. CTL, control.