LIN28B-mediated PI3K/AKT pathway activation promotes metastasis in colorectal cancer models

Abstract

Colorectal cancer (CRC) remains a leading cause of cancer death because of metastatic spread. LIN28B is overexpressed in 30% of CRCs and promotes metastasis, yet its mechanisms remain unclear. In this study, we genetically modified CRC cell lines to overexpress LIN28B, resulting in enhanced PI3K/AKT pathway activation and liver metastasis in mice. We developed genetically modified mouse models with constitutively active Pik3ca that form intestinal tumors progressing to liver metastases with an intact immune system, addressing the limitations of previous Pik3ca-mutant models, including long tumor latency, mixed histology, and lack of distant metastases. The PI3Kα-specific inhibitor alpelisib reduced migration and invasion in vitro and metastasis in vivo. We present a comprehensive analysis of vertical inhibition of the PI3K/AKT pathway in CRC using the FDA-approved drugs alpelisib and capivasertib (an AKT inhibitor) in combination with LY2584702 (a ribosomal protein S6 kinase inhibitor) in CRC cell lines and mouse- and patient-derived organoids. Tissue microarrays from patients with CRC verified that LIN28B and PI3K/AKT pathway activation correlate with CRC progression. These findings highlight the critical role of the LIN28B-mediated PI3K/AKT pathway in CRC metastasis, the therapeutic potential of targeted inhibition, and the promise of patient-derived organoids in precision medicine in metastatic CRC.

Keywords: Colorectal cancer; Gastroenterology; Mouse models; Oncogenes; Oncology.

Figure 1. LIN28B expression in CRC cells activates the PI3K/AKT pathway and promotes liver metastasis.

(A) Western blot of LIN28B protein levels in LoVo and DLD-1 CRC cell lines with either EV or LIN28B overexpression vector (LIN28B^hi), normalized to GAPDH and LoVo EV (1-way ANOVA, mean ± SEM). (B) Experimental setup for in vivo colorectal liver metastasis assay. (C) Representative H&E and GFP images of liver sections from mice injected with CRC cells. Scale bars: 5 mm; scale bars for insets: 1 mm. (D) Proportion of mice that developed liver metastases (χ² test). (E) Quantification of the size of liver metastases in each group (1-way ANOVA, mean ± SEM). (F) GSEA showing hallmark pathways enriched in LoVo LIN28B^hi cells compared with EV cells (n = 3). (G) Western blot analysis of p-AKT (Ser473) and t-AKT in CRC cells (2-tailed Student’s unpaired t test, mean ± SEM). (H) Quantification of phosphorylated protein targets involved in the PI3K/AKT pathway in LIN28B^hi cells relative to EV cells as measured by AKT pathway phosphorylation array (2-tailed Student’s unpaired t test between EV and LIN28B^hi for each protein, mean ± SEM). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 2. Genetic activation of the PI3K/AKT pathway enhances organoid growth ex vivo and induces colonic crypt hyperplasia in vivo.

(A) Schematic representation of the genetic cross to generate Vil^Cre R26^Pik3ca mice on a C57BL/6J background. (B) Representative bright-field and GFP images of colonic organoids derived from R26^WT/WT, R26^Pik3ca/WT, and R26^{Pik3ca/Pik3ca} mice cultured for 5 days. Scale bars: 500 μm (n = 3). (C) Immunoblot quantification of p-AKT (Ser473) levels relative to GAPDH in colonic organoids derived from R26^WT/WT, R26^Pik3ca/WT, and R26^{Pik3ca/Pik3ca} mice, normalized to R26^WT/WT (n = 3; 1-way ANOVA, mean ± SEM). (D) Quantification of growth of colonic organoids from R26^WT/WT, R26^Pik3ca/WT, and R26^{Pik3ca/Pik3ca} mice, showing percentage increase in initial area (n = 3 and 4; 2-way ANOVA, mean ± SEM). (E) Quantification of the number of colonic organoids per well on day 3 of culture (1-way ANOVA, mean ± SEM). (F) Representative immunofluorescence and IHC images of colonic tissues from R26^WT/WT, R26^Pik3ca/WT, and R26^{Pik3ca/Pik3ca} mice, showing Pik3ca-GFP, p-AKT (Ser473), Ki67, and H&E staining. Scale bars: 100 μm. (G) Quantification of p-AKT–stained area per section, Ki67-positive cells per crypt, and crypt length in colonic tissues from R26^WT/WT, R26^Pik3ca/WT, and R26^{Pik3ca/Pik3ca} mice. Crypt length was measured every 100 μm along the length of the colon (1-way ANOVA, mean ± SEM). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 3. Genetic activation of the PI3K/AKT pathway promotes tumorigenesis, tumor invasiveness, and liver metastasis in a mouse model of CRC.

(A) Kaplan-Meier survival curves of Vil^Cre and Vil^CreERT2 mice with either R26^WT/WT, R26^Pik3ca/WT, or R26^{Pik3ca/Pik3ca} genotype (n = 7, 4, and 25 for Vil^Cre, n = 5, 4, and 11 for Vil^CreERT; log-rank test). (B) Representative H&E images of the colon and SI tissues from mice with R26^{Pik3ca/Pik3ca} genotype. Dashed lines outline the tumors. Scale bars: 100 μm. (C) Proportion of mice with noninvasive adenomas and invasive adenocarcinomas in the colon and SI from Vil^Cre and Vil^CreERT2 mice with either R26^WT/WT or R26^{Pik3ca/Pik3ca} genotype (Fisher’s exact test). (D) Representative H&E, CDX2, and Alcian blue staining of a liver metastasis from a 28-week-old Vil^Cre R26^{Pik3ca/Pik3ca} mouse. Scale bars: 1 mm; scale bars for insets: 100 μm. (E) Kaplan-Meier survival curves of Vil^Cre and Vil^CreERT2 mice treated with AOM and tamoxifen (n = 8–14 for Vil^Cre, n = 10–12 for Vil^CreERT; log-rank test). (F and G) Representative H&E images of colon, SI, and liver tissues from Vil^Cre (F) and Vil^CreERT2 (G) mice treated with AOM. Dashed lines outline the tumors. Scale bars: 100 μm. (H) Proportion of mice with noninvasive adenomas and invasive adenocarcinomas in the colon and SI, and liver metastases from mice treated with AOM (χ² test between R26^WT/WT and either R26^Pik3ca/WT or R26^{Pik3ca/Pik3ca} genotype). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 4. Alpelisib impairs LIN28B-induced cell migration and invasion and inhibits PI3Kα-induced organoid growth.

(A) Western blot analysis of p-AKT (Ser473) and t-AKT in CRC cells (1-way ANOVA, mean ± SEM). (B) Colony formation assay of LIN28B^hi CRC cells treated with 5 μM alpelisib. Scale bars: 500 μm (1-way ANOVA, mean ± SEM). (C) Wound healing assay showing cell migration of CRC cells treated with 5 μM alpelisib at 0 hours. Scale bars: 500 μm (n = 4; 2-way ANOVA, mean ± SEM). (D) Transwell ECM invasion assay of CRC cells treated with 5 or 10 μM alpelisib. Scale bars: 1 mm (1-way ANOVA, mean ± SEM). (E) Representative bright-field images of colonic organoids derived from Vil^Cre mice treated with 5 μM alpelisib every 2 days for 5 days. Scale bars: 500 μm. (F) Western blot analysis of LIN28B, p-AKT (Ser473), and t-AKT in colonic organoids derived from Vil^Cre mice treated with 5 μM alpelisib (2-tailed Student’s unpaired t test within each group, mean ± SEM). (G) Quantification of growth of colonic organoids derived from Vil^Cre mice treated with 5 μM alpelisib (n = 3–5; 2-way ANOVA, mean ± SEM; significance for day 5 is shown with asterisks; the asterisk above the data point signifies significance when compared with the R26^WT/WT control group). *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 5. Alpelisib inhibits colorectal liver metastasis formation in mice.

(A) Experimental setup for investigating the effect of alpelisib on colorectal liver metastasis formation. (B) Representative images of liver tissues from mice injected with CRC cells and treated with vehicle or alpelisib. Gross liver morphology with black arrows denoting liver metastases (top), GFP fluorescence indicating liver metastases from LIN28B^hi-GFP CRC cells (middle), and H&E staining with black arrows denoting liver metastases (bottom) are shown. Scale bars: 5 mm. (C) Weight change of mice over the course of the experiment, expressed as percentage of initial weight. Dashed lines, individual mice; solid line, average of all mice in group (n = 7 and 10). (D) Quantification of liver weight (1-way ANOVA, mean ± SEM). (E) Proportion of mice with liver metastases in each group (χ² test). The dataset for the control groups in this graph is the same as the data reported in Figure 1D. (F) Quantification of the area of liver metastases in each group (1-way ANOVA, mean ± SEM). (G) Experimental setup in which Vil^CreERT R26^{Pik3ca/Pik3ca} mice were treated with AOM to induce tumor formation, followed by tamoxifen, and subsequently treated with 25 μg/g alpelisib after primary tumors had formed. (H) Proportion of Vil^CreERT R26^{Pik3ca/Pik3ca} mice with tumors in the colon and SI and liver metastases (Fisher’s exact test). (I) Representative H&E-stained images of colon, SI, and liver tissues from Vil^CreERT R26^{Pik3ca/Pik3ca} mice treated with alpelisib. Scale bars: 1 mm; scale bars for insets: 100 μm. *P < 0.05.

Figure 6. Pharmacologic inhibition of the S6K/RPS6 axis suppresses LIN28B-driven cell migration and invasion in CRC cells.

(A) GSEA from RNA-Seq showing hallmark pathways enriched in LIN28B^hi cells compared with LIN28B^hi cells treated with 5 μM alpelisib. (B) GSEA from RNA-Seq showing hallmark pathways enriched in liver metastasis compared with matched primary tumors in patients with CRC (GSE50760). (C) Quantification of phosphorylated protein targets involved in the PI3K/AKT pathway in EV, LIN28B^hi, and LIN28B^hi cells treated with 5 μM alpelisib (1-way ANOVA for each protein target, mean ± SEM). (D) Western blot analysis of LIN28B, p-AKT (Ser473), p-S6K (Thr389/412), total S6K, p-RPS6 (Ser235/236), t-RPS6, and GAPDH in CRC cells treated with 5 μM alpelisib (alp, alpelisib; 1-way ANOVA, mean ± SEM). (E) Representative IHC images of p-S6K (Thr389/412) and p-RPS6 (Ser235/236) in colonic tissues from Vil^Cre mice. Scale bars: 100 μm. (F) Western blot analysis of p-RPS6, t-RPS6, and GAPDH in CRC cells treated with varying concentrations of LY2584702 (S6K inhibitor) (1-way ANOVA, mean ± SEM). (G) Wound healing assay showing cell migration of LIN28B^hi CRC cells (n = 3; 2-way ANOVA, mean ± SEM). Scale bars: 500 μm. (H) Transwell ECM invasion assay of LIN28B^hi CRC cells (1-way ANOVA, mean ± SEM). Scale bars: 1 mm. *P < 0.05, **P < 0.01, ****P < 0.0001.

Figure 7. Pharmacologic inhibition of PI3Kα and S6K impairs the growth of CRC PDOs.

(A) Heatmap showing top 200 CRC mutations in Catalogue of Somatic Mutations in Cancer (COSMIC) cancer genes in PDO lines identified by WES. Only genes with mutations are shown. Mutation types are color coded as indicated in the legend (T, primary tumor; met, liver metastasis). (B) Heatmap showing mutations in PI3K/AKT pathway genes in PDO lines identified by WES. Refer to the list of PI3K/AKT pathway genes used in Supplemental Methods. Only genes with mutations are shown. (C) Quantification of Western blot analysis of p-AKT (Ser473), phosphorylated mTOR (p-MTOR) (Ser2448), and p-RPS6 (Ser235/236) in PDO lines (1-way ANOVA, mean ± SEM). (D and E) Representative bright-field images (D) and growth curves (E) of PDOs treated with the inhibitors every other day for 8 days. Scale bars: 100 μm (n = 3; 2-way ANOVA, mean ± SEM). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 8. PI3K/S6K signaling correlates with disease progression in CRC patient samples.

(A) Representative IHC images of LIN28B, p-AKT (Ser473), p-S6K (Thr389/412), and p-RPS6 (Ser235/236) in normal adjacent colon tissue, primary colon tumor, and liver metastases from 60 patients with CRC. Scale bars: 100 μm; scale bars for insets: 10 μm. (B) Quantification of IHC staining scores for LIN28B, p-AKT, p-S6K, and p-RPS6 (n = 60; 1-way ANOVA, mean ± SEM). (C) T-distributed stochastic neighbor embedding (t-SNE) plots showing the expression of PIK3CA in all epithelial cells (top) and in T4 stage tumor cells (bottom) from the Human Colon Cancer Atlas single-cell sequencing dataset (c295) comprising 371,223 cells. (D) Dot plot showing scaled mean expression and percentage of cells expressing PIK3CA, MTOR, and RPS6KB1 across different cell clusters (normal colonic epithelial and tumor cells) identified in the Human Colon Cancer Atlas dataset. cE, colonic epithelium. ***P < 0.001, ****P < 0.0001.