Targeting Lin28 axis enhances glypican-3-CAR T cell efficacy against hepatic tumor initiating cell population

Abstract

Overexpression of Lin28 is detected in various cancers with involvement in the self-renewal process and cancer stem cell generation. In the present study, we evaluated how the Lin28 axis plays an immune-protective role for tumor-initiating cancer cells in hepatocellular carcinoma (HCC). Our result using HCC patient samples showed a positive correlation between indoleamine 2,3-dioxygenase-1 (IDO1), a kynurenine-producing enzyme with effects on tumor immune escape, and Lin28B. Using in silico prediction, we identified a Sox2/Oct4 transcriptional motif acting as an enhancer for IDO1. Knockdown of Lin28B reduced Sox2/Oct4 and downregulated IDO1 in tumor-initiating hepatic cancer cells. We further observed that inhibition of Lin28 by a small-molecule inhibitor (C1632) suppressed IDO1 expression. Suppression of IDO1 resulted in a decline in kynurenine production from tumor-initiating cells. Inhibition of the Lin28 axis also impaired PD-L1 expression in HCC cells. Consequently, modulating Lin28B enhanced in vitro cytotoxicity of glypican-3 (GPC3)-chimeric antigen receptor (CAR) T and NK cells. Next, we observed that GPC3-CAR T cell treatment together with C1632 in a HCC xenograft mouse model led to enhanced anti-tumor activity. In conclusion, our results suggest that inhibition of Lin28B reduces IDO1 and PD-L1 expression and enhances immunotherapeutic potential of GPC3-CART cells against HCC.

Keywords: CAR T cell; IDO1; Lin28B; PD-L1; glypican-3; hepatocellular carcinoma.

Graphical abstract

Figure 1

Overexpression of Lin28 axis and IDO1 in HCC (A and B) Comparison of Lin28A, Lin28B, Sox2, Oct4,IDO1, IDO2, and TDO2 transcripts analyzed using qRT-PCR in HCC (n = 12) and corresponding non-tumor tissue (n = 12) from liver specimens. (C) Pearson’s correlation plot of Lin28B and IDO1 transcript in the same liver specimens (r = 0.7714, p = 0.0006). (D and E) Expression of Lin28B and IDO1 mRNA in HCC (n = 200) and non-tumor (n = 50) samples obtained from The Cancer Genome Atlas (TCGA) database. ∗p < 0.05 and ∗∗p < 0.005. Immunohistochemistry of liver tissues for Lin28B and IDO1 expression in tissue microarray containing 40 cases of hepatocellular carcinoma, 17 normal liver tissues, and 13 adjacent non-tumor tissues, single core per case (BC03116a-Biomax), are shown. Tissue slide was treated with xylene for de-paraffinization. After antigen retrieval, slide was treated with LIN28B (Thermo Fisher) and IDO1 (Santa Cruz) antibodies and stained. (F) Tabular form of tissue microarray result is shown indicating the number of samples positive for Lin28B and IDO1 expression. Statistical analysis was performed using chi-square test with Yate’s correction. (G) An illustration of normal liver tissue and HCC with Lin28B (stained green) and IDO1 (stained brown) protein expressions are shown. The images were taken at 20× magnification.

Figure 2

Lin28 axis and IDO1 expression in HCC cell lines (A–D) Expression of Lin28B, Sox2, Oct4, and IDO1 mRNAs from qRT-PCR analysis from six different HCC cell lines (HepG2, THH, PLC/PRF5, HLF, Huh7, and Hep3B) with or without INF-γ treatment. The results are presented as the mean and SD from at least three independent experiments. ∗p < 0.05 and ∗∗p < 0.005. (E) Percentage of CD133⁺CD44⁺ cell population present in different HCC cell lines analyzed using flow cytometry.

Figure 3

Sox2 and Oct4 promote IDO1 expression (A) In silico diagram of Sox2/Oct4 motif on the basis of positional weight matrix (top), predicted sequences of Sox2/Oct4 motif at the cis-regulatory region of IDO1 (middle), and mutated sequences of Sox2/Oct4 motif at the cis-regulatory region of IDO1 (bottom). (B–D) IDO1 promoter activity was measured using luciferase reporter assay after 48 h. For this, pCpGL-IDO1-promoter plasmid was co-transfected with Sox2-siRNA, Oct4-siRNA, or control siRNA in THH, Huh7, and Hep3B cells with or without IFN-γ stimulation. (E–G) IDO1 promoter activity was measured using luciferase reporter assay after 48 h of transfection with either wild-type or mutated pCpGL-IDO1-promo plasmid in THH, Huh7, and Hep3B cells with or without INF-γ stimulation. (H) IDO1 mRNA expression was analyzed using qRT-PCR after transfection of Sox2-siRNA Oct4-siRNA, and control siRNA in THH, Huh7, and Hep3B cells. The results are presented as the mean and SD from at least three independent experiments. ∗p < 0.05, ∗∗p < 0.005, and ∗∗∗p < 0.0005. (I–K) IDO1 expression was analyzed using western blot after transfection of Sox2-siRNA, Oct4-siRNA, or control siRNA in THH, Huh7, and Hep3B stimulated with INF-γ. The blots were reprobed with antibody to actin as an internal control for comparison. (L) Schematic representation of Lin28 mediated IDO1 expression.

Figure 4

Pharmacological inhibition of Lin28 lowers kynurenine production (A and B) Huh7 and Hep3B cell proliferation was determined by MTS assay at different time points for 96 h after C1632 treatment in a dose-dependent manner. (C–F) Expression of Sox2, Oct4, and IDO1 were analyzed in Huh7 and Hep3B cells using western blot after C1632 treatment in a concentration-dependent manner. The blots were reprobed with antibody to actin as an internal control for comparison. (G and H) Cell proliferation was determined by MTS assay at different time points for 96 h after C1632 treatment in a concentration-dependent manner from CD133⁻CD44⁻ and CD133⁺CD44⁺ cell populations isolated from THH. (I–L) Expression of Sox2, Oct4, and IDO1 were analyzed using western blot after C1632 treatment with increasing doses in CD133⁻CD44⁻ and CD133⁺CD44⁺ cell populations. The blots were reprobed with antibody to actin as an internal control for comparison. (M) Extra-cellular kynurenine production from the culture supernatant of Huh7, Hep3B, CD133⁻CD44⁻, and CD133⁺CD44⁺ cells were measured at 72 h following treatment of C1632 in a dose-dependent manner. The results were presented as the mean and SD from at least three independent experiments. ∗p < 0.05, ∗∗p < 0.005, and ∗∗∗p < 0.0005.

Figure 5

Lin28 axis regulates PD-L1 expression (A) Comparison of PD-L1 transcripts analyzed using qRT-PCR in HCC tissue (n = 12) and corresponding non-tumorous tissue (n = 12) from liver biopsy specimens. (B) Pearson’s correlation plot of Lin28B and PD-L1 transcript in the same liver biopsy specimens (r = 0.6716, p = 0.0009). (C) Expression of PD-L1 mRNA in HCC (n = 200) and corresponding non-tumorous (n = 50) samples obtained from The Cancer Genome Atlas (TCGA) database. (D) Expression of PD-L1 mRNA was analyzed using qRT-PCR from six different HCC cells HepG2, THH, PLC/PRF5, HLF, Huh7, and Hep3B. (E–I) Expression of PD-L1was determined using qRT-PCR and western blot after transfection of Lin28B-siRNA, Sox2-siRNA, Oct4-siRNA, and control siRNA in Huh7, Hep3B, CD133⁻CD44⁻, and CD133⁺CD44⁺ cells. (J–M) PD-L1 expression was analyzed using western blot after C1632 inhibitor (10 μM) treatment with increasing doses in Huh7, Hep3B, CD133⁻CD44⁻, and CD133⁺CD44⁺ cells. The blots were reprobed with antibody to actin as an internal control for comparison. The results were presented as the mean and SD from at least three independent experiments. ∗p < 0.05, ∗∗p < 0.005, and ∗∗∗p < 0.0005.

Figure 6

Lin28 inhibitor increases tumor cell-killing activity of immune cells (A) Glypican-3 (GPC3) expression was analyzed using western blot after C1632 inhibitor treatment (10 μM) of Huh7, Hep3B, CD133⁻CD44⁻, and CD133⁺CD44⁺cells. GPC3 expression of THH and HepG2 cells were also included. The blots were reprobed with antibody to actin as an internal control for comparison. (B–E) GPC3-targeted CAR T cell-mediated killing activity was determined using fluorometric quantification of Calcein AM release after 3 h exposure of GPC3-CAR T or control CAR T cells at variable effector-to-target ratios against Huh7, Hep3B, CD133⁻CD44⁻, and CD133⁺CD44⁺ cells with or without C1632 inhibitor (10 μM) treatment. (F and G) A similar assessment was performed by colorimetric quantification of LDH release after 3 h exposure of GPC3-CAR T or control CAR T cells at a 2:1 effector-to-target ratio against the hepatic tumor-initiating cells with or without C1632 inhibitor (10 μM) treatment. (H) Tumor cell-killing activity of NK3.3 human natural killer cells after 3 h exposure at 2:1 effector-to-target ratio against hepatic tumor-initiating cells with or without C1632 inhibitor (10 μM) assessed by fluorometric quantification of Calcein AM release. The results were presented as the mean and SD from at least three independent experiments. ∗p < 0.05, ∗∗p < 0.005, and ∗∗∗p < 0.0005. (I) Schematic representation showed inhibition of Lin28 axis enhances immune cell-mediated hepatic tumor suppression.

Figure 7

Lin28 inhibitor improves in vivo GPC3-CAR T cell anti-tumor activity (A) Schematic representation of the in vivo experimental procedure. NSG mice were injected intraperitoneally (i.p.) with 10⁵ THH derived CD133⁺CD44⁺-Luc cells followed by the intravenous (i.v.) injection of 10⁶ GPC3-CAR or control CAR T cells in the second week after establishment of the HCC xenograft. C1632 inhibitor (10 mg/kg) was introduced three times before and after the CAR T cell injection; tumor growth was checked weekly using IVIS. (B) Bioluminescence imaging was monitored for the different indicated groups of mice over the experimental time course. The mouse not bearing tumor is covered by a gray box. (C) Tumor bioluminescence as photon count is represented by each mice group at indicated time points. (D) Kynurenine level was measured from the peritoneal fluid of each mouse group. (E) Human INF-γ secretion was measured using ELISA from the peritoneal fluid of each mouse group. The results are presented as the mean with SD. ∗p < 0.05, ∗∗p < 0.005, and ∗∗∗p < 0.0005.