Fig 2: WSX1 relieves inhibition of GSK3ß activity through impeding the PI3Kd/AKT signaling pathway.a Effect of WSX1 overexpression on protein levels of p-AKTSer473, AKT, p-TSC2Thr1462, and TSC2 in HCC cells determined by immunoblotting. TSC2 is directly phosphorylated at Thr1462 by AKT; thus p-TSC2Thr1462 levels indirectly reflect AKT activity. b Impact of AKT reactivation on WSX1-mediated PD-L1 downregulation detected by flow cytometry. SNU449 and SNU475 cells were transfected with WSX1 alone or co-transfected with membrane-bound myr-AKT. Statistical analysis results are shown on the right (n = 5 independent experiments, P < 0.0001 in both SNU449 and SNU475 cells). c Impact of myr-AKT transfection on protein levels of p-AKTSer473, total AKT, and PD-L1 in 449WSX1 and 475 WSX1 cells. d Immunoblotting analysis of expression of WSX1, PD-L1, and p-AKTSer473 in whole liver lysates obtained from FVB/NJ mice injected with NRAS/AKT oncogenes with or without WSX1 (left, C1–C4 represent independent samples from “oncogene” group, T1–T4 are independent samples from “oncogene + WSX1” group). Statistical analysis of correlations of WSX1 with PD-L1 and p-AKTSer473 expression in mouse livers (right). e Effect of WSX1 on expression of PTEN, PI3K-p85, PI3K-p110a, and PI3K-p110d. f Impact of PIK3CD overexpression on WSX1-mediated PD-L1 reduction (left). PIK3CD encodes PI3K-p110d, which is the key component of PI3Kd. Reintroduction of PIK3CD impaired WSX1-induced PD-L1 reduction in both SNU449 (n = 5 independent experiments, P = 0.0001) and SNU475 cells (P = 0.0013). g Effect of WSX1 on PIK3CD mRNA levels in SNU449 (n = 3 independent experiments P = 0.0139) and SNU475 HCC cells (P = 0.0006). h A proposed model illustrating that WSX1 facilitates antitumor immune surveillance through inhibiting the PI3Kd/AKT/GSK3ß/PD-L1 signaling pathway. Under physiological conditions, highly expressed WSX1 in hepatocytes transcriptionally downregulates PIK3d, thereby reducing AKT activation and subsequently liberating GSK3ß kinase activity from inhibition by AKT, leading to boosted GSK3ß-mediated PD-L1 degradation. Without excessive PD-L1 expression on tumor cells, effector CD8+ T cells maximize their killing effect, resulting in inhibition of HCC development (Left). Otherwise, lack of WSX1 results in uncontrolled neoplastic PD-L1 expression and, ultimately, tumor immune evasion (Right). i Schematic diagram of the interactions among WSX1, PI3Kd, AKT, GSK3ß, and PD-L1. Quantitative data are presented as mean ± SD and were analyzed by one-way ANOVA or Student t test. Tukey-Kramer multiple comparison test was used for pairwise comparisons in the ANOVA analysis. Correlation analyses were performed by Pearson correlation test. Unless otherwise noted, data and images shown are representative of 3 independent experiments. All statistical tests were two-sided. *P < 0.05, **P < 0.01, ***P < 0.001. myr-AKT myristoylated AKT. The gating strategy for sorting PD-L1+ HCC cells is shown in supplementary Fig. 8b. Source data are provided as a Source Data file.

Fig 3: BFCN respond to insulin and NGF by inducing c-Fos expression, Glut2 translocation, and glucose uptake (a-a1) Triple immunolabeling (a, 40×) and high magnification (a1, × 100) of untreated cholinergic neurons (DIV10) using specific anti-IR (pseudocolor as purple), anti-IRS1 (red), and anti-ChAT (pseudocolor as green) antibodies clearly confirm the co-expression of IR and IRS1 in primary cholinergic neurons. Representative images are from at least 30 neurons. (b–e) Representative WB of pIR Y1150/1151(b, c), pIRS1Y608 (b, d), and pAKT (b, e) on extracts from untreated (CTR), insulin-treated (INS, 10 nM, 30'), and NGF-treated (100 ng/ml, 30') neurons are shown. The results are reported as a percentage of the control septal neurons. As shown, the levels of pIRY1150/1151 (531.1 ± 112.4% of CTR; **p < 0.01; Fig. 2c), pIRS1Y608 (296.7 ± 35.5% of CTR; **p < 0.01; Fig. 2d), and pAKT (344.1 ± 38.8; **p < 0.01; Fig. 2e) are increased in insulin-treated cholinergic neurons, as compared to unstimulated neurons (CTR). After NGF stimulation the levels of pIRY1150/1151(274.7 ± 56.5% of CTR; *p < 0.05; Fig. 2c), pIRS1Y608 (207.1 ± 18.1% of CTR; *p < 0.05; Fig. 2d) and AKT (248.9 ± 14.5% of CTR; **p < 0.01; Fig. 2e) were increased, as compared to unstimulated neurons (CTR). (f–i) The c-Fos nuclear staining and the GluTracker fluorescence analyses in untreated (CTR), insulin-treated (INS, 10 nM, 30'), and NGF-treated (NGF, 100 ng/ml, 30') cholinergic neurons are shown. (f-g) Representative immunofluorescence images (f) and bar graphs (g) of c-Fos expression in cholinergic neurons following insulin and NGF treatments. Neurons were fixed, permeabilized and immunostained for c-Fos (green) and nuclei counterstained with DAPI. (g) The quantitative analysis of the cFos-positive nuclei/field (× 20 magnification) is reported in the graph as a percentage of the value in untreated neurons. Insulin and NGF stimulation of cholinergic neurons increased the number of c-Fos positive nuclei per field (INS, 222.7 ± 20.1% of CTR; **p < 0.01; NGF, 383.3 ± 74.1% of CTR, **p < 0.01). (h) Representative GluTracker immunofluorescences in cell bodies (arrows) and neurites (arrowheads) of untreated (CTR), insulin-treated (INS, 10 nM, 30'), and NGF-treated (NGF, 100 ng/ml, 30') cholinergic neurons are shown. (i) The GluTracker fluorescence intensity/field was measured and reported as percentage of CTR. Insulin and NGF significantly enhanced glucose uptake in cholinergic neurons (INS, 280.3 ± 6.7% of CTR; **p < 0.01 and NGF, 183.2 ± 33.1% of CTR; *p < 0.05). (J) Representative confocal microscopy images of glucose transporter 2 (GLUT2; green) of untreated (CTR), insulin-treated (INS, 10 nM, 30'), and NGF-treated (NGF, 100 ng/ml, 30') cholinergic neurons. As shown, both insulin and NGF treatments elevated plasma membrane localization of GLUT2 in the cell body (arrows) and axons/neurites (arrowhead). Scale Bars: a, a’, f, h, j = 25 µM

Fig 4: Combination of ACLY inhibitor and IGF1R inhibitor dramatically decreases HOXA13-mediated CRC metastasis.A The SW480-HOXA13 cells were respectively treated with ACLY inhibitor ETC-1002 (100 µM, 12 h), IGF1R inhibitor Linsitinib (10 µM, 6 h) or both two agents. The protein expression levels of p-AMPKa, total AMPKa, p-IGF1R, and total IGF1R were detected by western blotting. B Transwell assays showed the migratory and invasive capabilities of SW480-HOXA13 cells after treatment with ACLY inhibitor ETC-1002 (100 µM, 12 h), IGF1R inhibitor Linsitinib (10 µM, 6 h) or the combination of both two agents. C The diagram of agent treatment in four randomly assigned groups of nude mice. These mice were treated with vehicle, ETC-1002 (30 mg/kg), Linsitinib (50 mg/kg) or combined treatment 1 week after injection of SW480-HOXA13 cells. D In vivo liver metastatic assays revealed that combined administration of ETC-1002 and Linsitinib significantly suppress HOXA13-mediated CRC metastasis. Four stable cell lines were injected into the spleens of nude mice (n = 10 mice per group). Representative bioluminescent imaging, intensity of bioluminescence signals, incidence of liver metastasis, liver tissues’ HE staining, number of metastatic liver nodules, and overall survival times in each group were shown. The scale bars represent 1 mm (low magnification) and 100 µm (high magnification). E In vivo lung metastatic assays revealed that combined administration of ETC-1002 and Linsitinib significantly suppress HOXA13-mediated CRC metastasis. Four stable cell lines were injected into the tail veins of nude mice (n = 10 mice per group). Representative bioluminescent imaging, intensity of bioluminescence signals, number of metastatic lung nodules, overall survival times, lung tissues’ HE staining in each group, and the incidence of lung metastasis were shown. The scale bars represent 1 mm (low magnification) and 100 µm (high magnification). F A schematic diagram for the role of IGF1-HOXA13-IGF1R positive feedback loop in CRC metastasis. IGF1-IGF1R pathway upregulates HOXA13 via PI3K/AKT/HIF1a pathway. HOXA13 promotes CRC metastasis through upregulating ACLY and IGF1R expression. Combined administration of ACLY inhibitor (ETC-1002) and IGF1R inhibitor (Linsitinib) disrupts the IGF1-HOXA13-IGF1R loop and inhibits CRC metastasis. *P < 0.05, **P < 0.01. Data are displayed as the mean ± s.d.

Fig 5: ATM promotes GLUT1 translocation and PKM2 activation in CAFs. a. CAFs were cultured in normoxia or hypoxia for 8 h and treated with or without KU60019 (5 µM), the membrane GLUT1 was shown by immunofluorescence staining. Scale bar, 50 µm. b. The engineered CAFs (CAF/ecto-WT and CAF/ecto-S490A) (as described at material and methods) were cultured in normoxic or hypoxic condition for 8 h and treated with or without KU60019 (5 µM), the membrane location of GLUT1 in CAFs was detected by immunofluorescence staining. Scale bar, 50 µm. c. Glucose consumption and lactate production of CAFs described in (b) and their parent CAFs were detected. (a, p < .01, normoxic CAF/ecto-S490A vs normoxic CAF/ecto-WT under DMSO treatment; b, p < .01, hypoxic CAF/ecto-WT vs normoxic CAF/ecto-WT; c, p > .05, hypoxic CAF/ecto-S490A vs normoxic CAF/ecto-S490A; d, p < .01, hypoxic CAF/ecto-WT under KU60019 vs DMSO. ANOVA). d. CAFs were cultured in normoxia or hypoxia for 8 h and treated with or without KU60019 (5 µM), protein levels of p-ATM, ATM, p-AKT, AKT and PKM2 were determined by Western blotting. e. Western blotting was used to measure the expressions of p-ATM, ATM, p-AKT, AKT and PKM2 in control CAFs (CAF/sh-NC) and CAFs with shRNAs targeting ATM (CAFs/shATM#1 and CAFs/shATM#2) under hypoxic condition. ß-Actin is the loading control in (D) and (E). f. Immunofluorescence staining of PKM2 in CAFs as described in (e). DAPI is specifically dyed for nucleus. Scale bar, 100 µm.