Metastasis is responsible for the majority of deaths of patients with cancer. However, mechanisms governing metastasis in colorectal cancer remain largely unknown. Here we investigated how colorectal cancer cells acquire metastatic potential using a novel mouse model of colorectal cancer that spontaneously develops liver metastasis, generated by introducing sporadic mutations of Ctnnb1, Kras, Trp53, and Smad4 (CKPS) genes. Proteomic analyses revealed elevated expression of colorectal cancer stem cell markers ALCAM (CD166) and PROM1 (CD133) in colorectal cancer cells from the metastatic model compared with those from a nonmetastatic model. Spleen-to-liver metastasis assays using colorectal cancer cells derived from the CKPS model (CKPS cells) demonstrated the functional importance of ALCAM and PROM1 in initiating metastasis. Genetic and pharmacologic analyses using CKPS cells in 2D and spheroid culture revealed that expression of ALCAM and PROM1 is regulated positively and negatively by the cAMP/PKA/CREB and TGFβ/SMAD4 pathways, respectively. Consistently, phospho-CREB was expressed in both primary and metastatic lesions of CKPS mice and patients with colorectal cancer, and knockout of CREB in CKPS cells reduced their spheroid-forming and metastasis-initiating abilities. Treatment with a CREB inhibitor potentiated the effect of irinotecan in suppressing liver metastasis by CKPS cells. These results reveal the essential roles of ALCAM and PROM1, as well as their upstream regulators, the cAMP/PKA/CREB and TGFβ/SMAD4 pathways, in maintaining the stemness and metastatic potential of colorectal cancer cells and indicate that CREB inhibition may be a potential therapeutic strategy against metastatic colorectal cancer.

Significance:

This study identifies signaling pathways essential for maintaining the stemness and metastatic potential of colorectal cancer cells and proposes CREB as a therapeutic target in metastatic colorectal cancer.

Colorectal cancer is one of the leading global causes of cancer-related deaths (1). Despite recent advances in surgical treatment and chemotherapy, patients with stage IV colorectal cancer with distant metastasis have worse prognosis than stage I to III patients (2), necessitating identification of metastatic mechanisms and therapeutic targets (3). A large-scale analysis using clinical colorectal cancer specimens demonstrated mutations leading to activation of the Wnt/β-catenin pathway or KRAS, as well as those leading to inactivation of the p53 or the TGFβ pathway, in metastatic colorectal cancer (4). The study also indicated a lack of newly acquired driver mutations in metastatic lesions compared with primary tumors, in most cases (4), as suggested for other types of cancer (5). Studies using transplantation of organoids derived from genetically-engineered mouse models (GEMM) of colorectal cancer also demonstrated that some combinations of three or four mutations frequently found in clinical colorectal cancer suffice to initiate liver metastasis (6). Consistently with the roles of nongenetic factors suggested by these findings, epithelial–mesenchymal transition (EMT; ref. 7), cancer stem cells (8), and tumor microenvironment or premetastatic niche (9, 10) have been implicated in metastasis (5). However, it remains mostly unclear what molecular mechanisms and signaling pathways are involved in such nongenetic alterations in metastatic colorectal cancer.

Here we present a novel mouse model of sporadic colorectal cancer that metastasizes to the liver, generated by introducing mutations of 4 colorectal cancer-related genes in small populations of intestinal epithelial cells (IEC), taking advantage of spontaneous activation of CreERT2 recombinase (11). Analyses of this model revealed that the colorectal cancer stem cell markers, ALCAM (CD166) and PROM1 (CD133), regulated by the TGFβ/SMAD4 and cAMP/Protein kinase A (PKA)/CREB pathways, are essential to maintain the plasticity and metastasis-initiating activity of colorectal cancer stem cells.

Colorectal cancer mouse models

Creation of cis-Apc/Smad4 mice and Ctnnb1flox(ex3) mice has been described previously (12, 13). KrasLSL-G12D (Stock No.: 008179; RRID: IMSR_JAX:008179), Trp53flox/flox (Stock No.: 008462; RRID: IMSR_JAX:008462), Smad4flox/flox (Stock No.: 017462; RRID: IMSR_JAX:017462), and Lgr5-EGFP-IRES-CreERT2 (Stock No.: 008875; RRID: IMSR_JAX:008875) were purchased from Jackson Laboratories. Villin-CreERT2 mice were obtained from Sylvie Robine (Institut Curie; ref. 14). These mice were crossed in various combinations to generate models used in this study. All mice were maintained on a C57BL/6 background. Mice were kept under specific pathogen-free conditions in a 12-hour dark/12-hour light cycle and fed a standard chow diet and water ad libitum. All animal experiments were conducted according to protocols approved by the Animal Care and Use Committee of Aichi Cancer Center Research Institute.

Tumor scoring, histologic analysis, and immunostaining

Mice were euthanized and their intestines were excised and opened longitudinally. Numbers of intestinal adenomas and adenocarcinomas, as well as liver metastases were counted under a dissection microscope. For irinotecan treatment experiments, liver metastasis was scored by the percentage of area occupied by metastatic lesions in the liver using Image J software (15), because individual metastatic foci were too numerous to count in control mice 28 days after transplantation of CKPS cells. Tumor tissues were fixed with 10% formalin overnight. 4-μm sections of paraffin-embedded samples were stained with hematoxylin and eosin for pathologic analysis. Invasion-front cells were defined as tumor cells that had invaded beyond the muscularis propria, and luminal-side tumor cells were defined as those in glandular structures facing the lumen. For clinical samples, we further defined middle-region cells as those that are neither in the luminal-side nor in the invasion front. Tissue sections of primary tumors and liver metastases of human colorectal cancer were obtained from surgical samples at Aichi Cancer Center Hospital with written informed consent. Detailed protocols for staining and scoring protocols are in the Supplementary Materials and Methods.

RNA sequencing and RT-qPCR analysis

RNA was extracted from primary intestinal tumors, adjacent normal tissues, and liver metastases from CKPS mice using RNeasy Kits, according to the manufacturer's instructions (74134; Qiagen), and submitted to AnnoRoad Gene Technology (http://www.annoroad.com) for library preparation and sequencing in 150-nt paired-end mode on an Illumina HiseqX platform. Details on RNA sequencing (RNA-seq) data filtering and gene quantification and on RT-qPCR analysis are shown in the Supplementary Materials and Methods.

Whole-exome sequencing

DNA was extracted and purified from primary intestinal tumors, adjacent normal tissues, and liver metastases in CKPS mice using DNeasy Blood & Tissue Kits (69504; Qiagen), and submitted to GENEWIZ (https://www.genewiz.com/en/) for exosome sequencing. Exons were pulled down from the purified DNA with an Agilent SureSelect Mouse All Exon Kit (5190–4641; Agilent) and sequenced using the HiseqX platform. Methods for identification of somatic mutations are in the Supplementary Materials and Methods.

Proteomic analysis

Three primary intestinal adenocarcinomas, three surrounding normal tissues, and three liver metastatic lesions were collected from each of three CKPS mice. Locally invasive intestinal adenocarcinomas and surrounding normal tissues were collected from 5 cis-Apc/Smad4 mice. Detailed methods for proteomic analysis are in the Supplementary Materials and Methods.

Establishment and culture of mouse colorectal cancer cell lines

CKPS and CKP cells were established from colons of respective compound mutant mice via organoid cultures. Detailed information on organoids with other genetic backgrounds and methods for 2D and spheroid cultures, drug treatment, cell proliferation assays, and matrigel invasion assays are in the Supplementary Materials and Methods.

Gene overexpression and knockout in cell lines

The Prkaca gene was transiently overexpressed by lipofection in CKPS cells. Overexpression of other genes or gene knockout with CISPR-Cas9 was introduced by lentivirus. Details on plasmids, lentivirus production, and guide RNA sequences are in the Supplementary Materials and Methods.

Western blot and antibody array analysis

Proteins extracted from cell lines were separated by SDS-PAGE and transferred to PVDF membranes. After incubation with primary antibody, membranes were incubated by HRP-labeled secondary antibody. Blots were detected by chemiluminescence. Details of reagents, antibodies, and antibody arrays are shown in the Supplementary Materials and Methods.

Survival analysis of patients with colorectal cancer using TCGA datasets

We obtained GDC TCGA Colon Cancer (COAD) and Rectal Cancer (READ) datasets from UCSC Xena (https://xena.ucsc.edu/) in June 2020; HTSeq-FPKM-UQ GDC Hub; Phenotype GDC Hub; survival data GDC Hub. Details are shown in the Supplementary Materials and Methods.

Data analysis, visualization, and enrichment analysis

R software was used for data analysis and visualization (R Core Team, 2020). Beeswarm plots were created using the beeswarm package. Principal component analysis (PCA) was performed using the prcomp function in the R statistical package. We used the same function in the e1071 package for linear support vector machines (SVM) with cost and gamma parameters set to default values. Details of enrichment analysis are shown in the Supplementary Materials and Methods.

Quantification and statistical analysis

Sample sizes of experiments were determined on the basis of our previous studies (16, 17). Random assignments and blind assessments were not performed. Statistical analysis was performed in R (18) with the EZR package (Saitama Medical Center, Jichi Medical University). Data were analyzed using two-tailed Student t tests or one-way ANOVA, and the post-hoc Tukey honestly-significant-difference test. A P value less than 0.05 was considered statistically significant.

Data availability

Datasets for RNA-sequence and whole-exome sequence in this study are available from the DNA Data Bank of Japan (DDBJ): Sequenced Read Archive under accession numbers PRJDB11268 (RNA sequence, https://ddbj.nig.ac.jp/resource/bioproject/PRJDB11268) and PRJDB11305 (whole-exome sequence, https://ddbj.nig.ac.jp/resource/bioproject/PRJDB11305).

Development of a novel mouse model of colorectal cancer that metastasizes to the liver

Loss-of-function mutations in the APC gene, found in more than 80% of patients with colorectal cancer, are thought to initiate colorectal carcinogenesis through activation of Wnt/β-catenin signaling. Gain-of-function mutations in the CTNNB1 gene, encoding β-catenin, are also found in ∼10% of patients with colorectal cancer. Because mutations in TP53, KRAS, and SMAD4 are frequently observed during malignant progression of colorectal cancer (19), we attempted to generate a mouse model of highly malignant intestinal adenocarcinomas by combining tamoxifen-inducible expression of stable β-catenin (Ctnnb1flox(ex3)/+) and constitutively active KRAS (KrasLSL-G12D/+), as well as tamoxifen-inducible deletion of p53 (Trp53flox/flox) and SMAD4 (Smad4flox/flox) in intestinal epithelial cells (IEC) using villin-CreERT2. We chose a conditional mutation of Ctnnb1 instead of Apc, anticipating that controlling tumor numbers might be easier with the Ctnnb1 mutation, which requires a single hit (heterozygous mutation) for tumor formation. When CKPS mice possessing all four mutations were treated with tamoxifen, within 8 weeks they became moribund due to numerous tumors in the intestines, even with one-time treatment; they did not live long enough to develop metastasis.

However, we noticed that some untreated CKPS mice became moribund at ∼15 weeks of age and found that they developed intestinal tumors, most likely due to rare ligand-independent, spontaneous activation of CreERT2 recombinase (Fig. 1A; ref. 11). We therefore left various compound mutant mice without tamoxifen treatment and found that all CKPS, CKP, CKS, and CPS mice, but not KPS mice developed intestinal tumors (Fig. 1B and C). In particular, 100% of CKPS mice developed at least one invasive intestinal adenocarcinoma and 23% of them showed liver metastasis (Fig. 1C). We confirmed that large primary tumors and liver metastases in CKPS mice had recombination of all four floxed alleles. On the other hand, some smaller noninvasive tumors did not contain recombinant Kras or Smad4 alleles, yet they all had recombinant Ctnnb1 alleles (Supplementary Figs. S1A–S1C). Histologic analysis indicated that CKPS mice without liver metastasis harbored adenomatous polyps and pedunculated hemorrhagic adenocarcinomas, whereas those with liver metastasis presented aggressive intestinal adenocarcinomas accompanied by ulcerated lesions with collapsed luminal sides and interstitial hypertrophy (Fig. 1D). Metastatic foci in the liver comprised intestinal adenocarcinoma cells with glandular structures and interstitial hypertrophy, similar to those in their primary tumors (Fig. 1D). Although most metastases formed in the liver, two CKPS mice presented lung metastases (Supplementary Fig. S2A).

Figure 1.

Spontaneous activation of CreERT2 led to generation of a spontaneous metastasis model of colorectal cancer. A, Schematic of the murine CKPS genotype. B, Representative photographs of intestinal adenocarcinomas and their liver metastases in CKPS mice. C, Pie charts showing the incidence of intestinal tumors in mice with different combinations of mutations, classified based on the most advanced tumors in each mouse (adenoma, locally invasive adenocarcinoma, or liver metastasis). D, Gross (top row) and microscopic (hematoxylin and eosin staining; bottom row) appearance of intestinal tumors in CKPS mice schematically showing their progression from adenomas (left) to invasive adenocarcinomas with hemorrhage (middle left), invasive adenocarcinomas that metastasize to the liver (middle right), and liver metastases (right). Scale bars, 1 mm for primary AC with metastasis and 500 μm for others. AC, adenocarcinoma. E, The number of intestinal tumors that developed in each mouse genotype (left), classification of all tumors in the mice examined (middle), and size distribution of adenomas and adenocarcinomas in each genotype (right). Ad, adenoma; AC, adenocarcinoma. Data are presented as beeswarm and box plots and assessed by one-way ANOVA and Tukey HSD test. *, P < 0.05. F, Kaplan–Meier survival curves of mice with different combinations of mutations.

Figure 1.

Spontaneous activation of CreERT2 led to generation of a spontaneous metastasis model of colorectal cancer. A, Schematic of the murine CKPS genotype. B, Representative photographs of intestinal adenocarcinomas and their liver metastases in CKPS mice. C, Pie charts showing the incidence of intestinal tumors in mice with different combinations of mutations, classified based on the most advanced tumors in each mouse (adenoma, locally invasive adenocarcinoma, or liver metastasis). D, Gross (top row) and microscopic (hematoxylin and eosin staining; bottom row) appearance of intestinal tumors in CKPS mice schematically showing their progression from adenomas (left) to invasive adenocarcinomas with hemorrhage (middle left), invasive adenocarcinomas that metastasize to the liver (middle right), and liver metastases (right). Scale bars, 1 mm for primary AC with metastasis and 500 μm for others. AC, adenocarcinoma. E, The number of intestinal tumors that developed in each mouse genotype (left), classification of all tumors in the mice examined (middle), and size distribution of adenomas and adenocarcinomas in each genotype (right). Ad, adenoma; AC, adenocarcinoma. Data are presented as beeswarm and box plots and assessed by one-way ANOVA and Tukey HSD test. *, P < 0.05. F, Kaplan–Meier survival curves of mice with different combinations of mutations.

Close modal

CKS and CPS mice developed high-frequency, invasive, intestinal adenocarcinomas, but none developed metastatic lesions (Fig. 1C and E; Supplementary Figs. S2B and S2C). On the other hand, CKP mice developed invasive adenocarcinomas less frequently, and 1 of 30 CKP mice showed liver metastasis (Fig. 1C and E; Supplementary Fig. S2D). Survival time was similar among compound mutant mice, except for CKP and KPS mice (Fig. 1F), which may reflect their lower incidence among invasive adenocarcinomas (Fig. 1C and E).

These results demonstrate establishment of a novel mouse model of colorectal cancer that spontaneously metastasizes to the liver. Our data suggest that activation of Wnt/β-catenin signaling initiates intestinal carcinogenesis and that additional mutations of Kras and Trp53 are required for liver metastasis. Comparison of CKPS and CKP mice suggests that Smad4 loss enhances invasion and metastasis of intestinal tumors.

Invasion-front cells of primary tumors in metastatic models show dedifferentiated phenotypes

To identify characteristic properties of metastatic intestinal adenocarcinomas in CKPS mice, we performed IHC analysis for known markers of proliferation and survival of colorectal cancer cells. The frequency of cleaved caspase 3-positive apoptotic cells was similar in luminal sides and invasion fronts of primary tumors and liver metastases (Supplementary Figs. S3A and S3B). However, the percentage of Ki67-positive proliferating cells per gland was high on luminal sides of tumors, but was sharply decreased in primary tumor invasion fronts and in liver metastases (Supplementary Figs. S3A and S3B). These results suggest that adenocarcinoma cells in invasion fronts or metastases may have different traits than those in the luminal side. To further explore this difference, we next examined expression of differentiation markers in CKPS tumors.

CDX2 is a transcription factor essential for differentiation of IECs during development and in adult intestinal homeostasis and is often used as an IEC marker (20). The CDX2 level is inversely correlated with poor prognosis in patients with colorectal cancer (21). We found that the frequency of CDX2-positive cells and the intensity of CDX2 staining are significantly decreased in tumor epithelial cells in primary tumor invasion fronts and liver metastases of CKPS/CKP mice, compared with those on luminal sides of primary tumors or of normal IECs (Fig. 2A and B), indicating reduced differentiation status of invasive tumor cells.

Figure 2.

Differentiation status is changed in primary tumor invasion fronts and metastases of CKPS and CKP mice compared with primary tumor luminal sides. A, Immunostaining of CDX2 in normal colon, primary adenocarcinoma (luminal side and invasion front), and liver metastases of CKPS mice, as well as in the normal duodenum, duodenal adenocarcinoma (luminal side and invasion front), and liver metastases of a CKP mouse. Scale bars, right bottom, 50 μm; others 100 μm. B, Percentages of CDX2-positive glands in luminal sides and invasion fronts of primary tumors in CKPS and CKP mice and in liver metastases. C, PCA analysis of RNA-seq data from normal intestinal tissues (Nrm), primary tumors (P), and liver metastases (Met) of CKPS mice. D, Enrichment analysis of transcription factors whose expression in RNA-seq analysis was increased more than two-fold in intestinal tumors of CKPS mice, compared with their normal intestinal tissues and maintained in liver metastases. A total of 73 transcription factors were analyzed with Metascape (customized to WikiPathway). Endoderm-related genes are listed on the right. E, Immunostaining of SOX17 in representative primary tumors and liver metastases of CKPS mice. Scale bars, 500 μm for top panels and 100 μm for bottom panels. F, Percentage of SOX17-positive cells in immunostaining of primary tumors (luminal and invasion front) and liver metastases of CKPS, CKP, CKS, and CPS mice. Data are presented as beeswarm and box plots (B and F), and assessed with one-way ANOVA and Tukey HSD test or with two-tailed Student t test. *, P < 0.05.

Figure 2.

Differentiation status is changed in primary tumor invasion fronts and metastases of CKPS and CKP mice compared with primary tumor luminal sides. A, Immunostaining of CDX2 in normal colon, primary adenocarcinoma (luminal side and invasion front), and liver metastases of CKPS mice, as well as in the normal duodenum, duodenal adenocarcinoma (luminal side and invasion front), and liver metastases of a CKP mouse. Scale bars, right bottom, 50 μm; others 100 μm. B, Percentages of CDX2-positive glands in luminal sides and invasion fronts of primary tumors in CKPS and CKP mice and in liver metastases. C, PCA analysis of RNA-seq data from normal intestinal tissues (Nrm), primary tumors (P), and liver metastases (Met) of CKPS mice. D, Enrichment analysis of transcription factors whose expression in RNA-seq analysis was increased more than two-fold in intestinal tumors of CKPS mice, compared with their normal intestinal tissues and maintained in liver metastases. A total of 73 transcription factors were analyzed with Metascape (customized to WikiPathway). Endoderm-related genes are listed on the right. E, Immunostaining of SOX17 in representative primary tumors and liver metastases of CKPS mice. Scale bars, 500 μm for top panels and 100 μm for bottom panels. F, Percentage of SOX17-positive cells in immunostaining of primary tumors (luminal and invasion front) and liver metastases of CKPS, CKP, CKS, and CPS mice. Data are presented as beeswarm and box plots (B and F), and assessed with one-way ANOVA and Tukey HSD test or with two-tailed Student t test. *, P < 0.05.

Close modal

Transcription factors are essential for cell differentiation and de-differentiation of many cell types, including cancer cells. Reduced expression of CDX2 prompted us to perform enrichment analysis of RNA-seq data for transcription factors, expression of which is elevated in CKPS primary tumors compared with normal intestinal tissues, and which is also maintained in liver metastases. These results indicate enrichment of those involved in mesodermal commitment and endoderm differentiation (Fig. 2C and D). Kaplan–Meier survival analysis of patients with colorectal cancer from the TCGA database showed that higher expression of SOX17, a definitive endoderm marker, was significantly correlated with shorter overall survival time for both patients with colon (COAD) and rectal (READ) cancer (Supplementary Fig. S3C). IHC analysis of surgically resected samples from patients with colon (COAD) and rectal (READ) cancer with liver metastasis demonstrated that SOX17-positive cells were found in both the primary tumors and liver metastases of the same patients (Supplementary Figs. S3D and S3E). Consistently, immunostaining revealed that SOX17 was highly expressed in adenocarcinoma cells at primary tumor invasion fronts and in liver metastases of CKP and CKPS mice, compared with those at the primary tumor luminal side (Fig. 2E and F). In contrast, SOX17 was barely expressed in nonmetastatic tumors of CPS and CKS mice (Fig. 2F). These results indicate that the regulation of gene expression differs between luminal sides and invasion fronts or metastases and suggest that intestinal adenocarcinoma cells may undergo dedifferentiation upon invasion and metastasis.

Stem cell-like populations are increased in primary tumor invasion fronts and in liver metastases

To better understand the mechanism of metastasis, we analyzed mutations in primary tumors and metastases of CKPS mice using exome sequencing. Exome sequence analysis demonstrated that metastatic lesions had not acquired meaningful translocations or mutations of cancer-related genes, or genes associated with cell proliferation, survival, or motility (Supplementary Figs. S4A and S4B). Although PCA analysis of RNA-seq data indicated no significant differences between primary tumors and liver metastases (Fig. 2C), proteomic data revealed a slight difference between them in the PC2 axis, whereas the difference in the PC1 axis appeared to represent a difference in tumor location: colon versus small intestine (Fig. 3A). UniProt annotation score analysis on the PC2 loading indicated that liver metastases show increases of nuclear-related proteins and decreases of mitochondrial proteins, compared with primary tumors (Fig. 3A). We next looked for individual proteins that were increased in liver metastases compared with primary tumors. Although some stem cell markers, including SLC12A2 (22) and ALCAM (CD166; refs. 23, 24), as well as the gut stem cell niche factor, REG4 (25), tended to be enriched in metastases, the increase was not statistically significant (Fig. 3B; Supplementary Figs. S5A and S5B). Only the level of PLA2G2A (26) was significantly higher in metastatic lesions compared with primary tumors (Supplementary Fig. S5A). On the other hand, levels of ALCAM and PROM1 (CD133) were significantly higher in primary tumors and liver metastases of CKPS mice than in nonmetastatic adenocarcinomas of cis-Apc/Smad4 mice (Fig. 3B).

Figure 3.

Expression of colorectal cancer stem cell markers is increased in primary tumor invasion fronts and liver metastases of CKPS mice. A, PCA analysis of proteomic data from normal intestinal tissues (Nrm), primary tumors (P), and liver metastases (Met) of CKPS mice (left) and enrichment analysis of the PC1 (middle) and PC2 (right) loading using UniProt annotation. B, Protein levels of ALCAM (CD166), PROM1 (CD133), and SLC12A2 in intestinal adenocarcinomas of cis-Apc/Smad4 mice (AS), primary tumors (Prm), and liver metastases (LM) of CKPS mice, based on proteomic analysis. Values represent changes compared with normal intestinal tissues. Data are presented as beeswarm and box plots and assessed with one-way ANOVA and Tukey HSD test. C, Immunostaining of PROM1 in representative primary tumors (luminal side and invasion front) and liver metastases of CKPS mice. Scale bars, 50 μm. D, Immunofluorescent staining of ALCAM (red) in normal colonic tissues, primary tumors, and liver metastases of CKPS mice. Epithelial cells were stained with anti-pan-cytokeratin (CK; green) and nuclei are stained with DAPI (blue). Scale bars, 100 μm for the top middle panel and 50 μm for other panels. Tu, tumor; N, normal mucosa; Inv, invasion front. E, Immunofluorescent staining of ALCAM (red) in primary intestinal tumors of CKP mice. Epithelial cells were stained with anti-pan-cytokeratin (green) and nuclei are stained with DAPI (blue). Scale bars, 50 μm for other panels. F, Quantification of the percentage of ALCAM-positive cells in intestinal tumors of CKPS and CKP mice. Data are presented as beeswarm and box plots. Statistical significance was assessed with two-tailed Student t test. *, P < 0.05.

Figure 3.

Expression of colorectal cancer stem cell markers is increased in primary tumor invasion fronts and liver metastases of CKPS mice. A, PCA analysis of proteomic data from normal intestinal tissues (Nrm), primary tumors (P), and liver metastases (Met) of CKPS mice (left) and enrichment analysis of the PC1 (middle) and PC2 (right) loading using UniProt annotation. B, Protein levels of ALCAM (CD166), PROM1 (CD133), and SLC12A2 in intestinal adenocarcinomas of cis-Apc/Smad4 mice (AS), primary tumors (Prm), and liver metastases (LM) of CKPS mice, based on proteomic analysis. Values represent changes compared with normal intestinal tissues. Data are presented as beeswarm and box plots and assessed with one-way ANOVA and Tukey HSD test. C, Immunostaining of PROM1 in representative primary tumors (luminal side and invasion front) and liver metastases of CKPS mice. Scale bars, 50 μm. D, Immunofluorescent staining of ALCAM (red) in normal colonic tissues, primary tumors, and liver metastases of CKPS mice. Epithelial cells were stained with anti-pan-cytokeratin (CK; green) and nuclei are stained with DAPI (blue). Scale bars, 100 μm for the top middle panel and 50 μm for other panels. Tu, tumor; N, normal mucosa; Inv, invasion front. E, Immunofluorescent staining of ALCAM (red) in primary intestinal tumors of CKP mice. Epithelial cells were stained with anti-pan-cytokeratin (green) and nuclei are stained with DAPI (blue). Scale bars, 50 μm for other panels. F, Quantification of the percentage of ALCAM-positive cells in intestinal tumors of CKPS and CKP mice. Data are presented as beeswarm and box plots. Statistical significance was assessed with two-tailed Student t test. *, P < 0.05.

Close modal

Immunostaining showed that PROM1 was expressed in adenocarcinoma cells on both the luminal side and the invasion front of primary tumors, as well as in liver metastases of CKPS mice (Fig. 3C). In contrast, ALCAM was weakly and only rarely expressed in adenocarcinoma cells on the luminal side of primary tumors, whereas it was strongly expressed in adenocarcinoma cells at primary tumor invasion fronts and in liver metastases of CKPS mice (Fig. 3D). Furthermore, little epithelial expression of ALCAM was observed at invasion fronts of nonmetastatic intestinal tumors in CKP mice (Fig. 3E and F). These findings suggested that ALCAM may be a marker of colorectal cancer stem cells with high metastatic potential.

We also determined expression of LGR5, a well-known intestinal stem cell marker, in primary intestinal tumors and liver metastases, using another version of CKPS mice driven by Lgr5-EGFP-IRES-CreERT2 (Lgr5:CKPS mice; ref. 27). Lgr5: CKPS mice developed metastatic intestinal adenocarcinomas similar to those in the original CKPS mice, although the frequency of metastasis was as low as 10%, possibly due to smaller populations of LGR5-positive cells in the crypt (Supplementary Figs. S6A–S6C). Immunostaining revealed some LGR5 (GFP)-positive cells in primary tumors of Lgr5: CKPS mice; however, we found few LGR5-positive cells in their metastatic lesions (Supplementary Figs. S6D and S6E). Taken together, these results indicate that some stem cell markers are highly expressed in CKPS intestinal adenocarcinomas, especially in primary tumor invasion fronts and in liver metastases, prompting us to further investigate their possible functions in metastasis.

SMAD4 negatively regulates expression of the stem cell markers, ALCAM and PROM1, and suppresses metastasis

To understand mechanisms underlying expression of ALCAM and PROM1 in CKPS tumors and their possible roles in metastasis, we first examined their expression during spheroid formation of CKPS-derived cells in 3D, free-floating culture (28), given their well-known roles as colorectal cancer stem cell markers (Fig. 4A). We found that CKP- and CKPS-derived cells were able to proliferate in both 2D and 3D cultures, forming spheroids in 3D, indicating that these cells have one of the in vitro characteristics of cancer stem cells (Fig. 4A and B). In contrast, C (Ctnnb1 single mutant)-, CP-, and CPS-derived cells failed to adhere to culture dishes and could not be cultured in 2D, whereas CKS-derived cells adhered to culture dishes, but could not be passaged further. These results suggest that loss of Smad4 is necessary for induction of colorectal cancer stem cell markers but may not be sufficient for efficient survival and proliferation of colorectal cancer cells in 2D or 3D cultures, and that p53 and Kras mutations are also required. We then examined expression of some intestinal stem cell markers in CKP, CKS, and CKPS cells 24 hours after the shift to spheroid culture and found that expression levels of PROM1 and ALCAM were lower in CKP cells than in other cells, suggesting that SMAD4 may regulate their expression (Fig. 4C). Consistently, protein and RNA levels of stem cell markers in spheroid culture were reduced by overexpressing SMAD4 in CKPS cells (Fig. 4D; Supplementary Figs. S7A and S7B). We found no significant difference between CKPS and CKP cells in their proliferation rates in 2D culture or in Matrigel invasion activity in vitro (Supplementary Fig. S7C).

Figure 4.

ALCAM and PROM1 positively and SMAD4 negatively regulate plasticity and metastasis-initiating activity of intestinal adenocarcinoma stem cells of CKPS mice. A, Schematic illustration for generating cell lines from CKP and CKPS organoids. B, Representative photomicrographs for 3D culture of organoid-derived cells. Scale bars, 100 μm. C, Western blots of PROM1 and ALCAM in 3D culture of organoid-derived cells. D, Western blots of PROM1 and ALCAM in CKPS cells, CKPS cells overexpressing SMAD4, and CKP cells in 2D culture or shifted to 3D for 24h. E, Schematic description of the liver metastasis model by allograft transplantation of 2D-cultured CKP or CKPS cells into spleens of C57BL/6 mice. F, Metastasis-initiating activity of CKP and CKPS cells. Left, the number of metastatic foci formed 20 days after injecting CKP (1 × 104 and 1 × 105) or CKPS (1 × 103 and 1 × 104) cells (n = 4). Data are shown as beeswarm and box plots, and statistical significance was assessed with two-tailed Student t test. Middle, gross appearance of liver metastatic foci. Scale bars, 10 mm. Right, images for hematoxylin and eosin (H&E) staining and PROM1 immunostaining in liver metastatic foci. Scale bars, 100 μm for hematoxylin and eosin staining and 50 μm for PROM1 staining. G, Western blots for verification of Prom1 and Alcam gene knockout by CRISPR in CKPS cells. N.C., negative control for CRISPR. 1 and 2, ID of knockout cell clones. H, Effect of Prom1 or Alcam knockout (KO) on the metastasis-initiating activity of CKPS cells. Left, the number of metastatic foci measured 20 days after injecting 1 × 104 N.C., Prom1 KO, or Alcam KO CKPS cells (n = 6). Middle, representative images of liver metastases. Scale bars, 10 mm. Right, hematoxylin and eosin staining of representative liver metastatic foci. Scale bars, 200 μm. Data are presented as beeswarm and box plots and assessed with one-way ANOVA and Tukey HSD test. I, The effect of Prom1 KO and Alcam KO on 3D-cultured CKPS cells evaluated by spheroid size. Representative micrographs after 15 days of culture (left) and size evaluation (right) after 9 and 15 days of culture. Scale bars, 200 μm for N.C. and 100 μm for Prom1 KO and Alcam KO. *, P < 0.05.

Figure 4.

ALCAM and PROM1 positively and SMAD4 negatively regulate plasticity and metastasis-initiating activity of intestinal adenocarcinoma stem cells of CKPS mice. A, Schematic illustration for generating cell lines from CKP and CKPS organoids. B, Representative photomicrographs for 3D culture of organoid-derived cells. Scale bars, 100 μm. C, Western blots of PROM1 and ALCAM in 3D culture of organoid-derived cells. D, Western blots of PROM1 and ALCAM in CKPS cells, CKPS cells overexpressing SMAD4, and CKP cells in 2D culture or shifted to 3D for 24h. E, Schematic description of the liver metastasis model by allograft transplantation of 2D-cultured CKP or CKPS cells into spleens of C57BL/6 mice. F, Metastasis-initiating activity of CKP and CKPS cells. Left, the number of metastatic foci formed 20 days after injecting CKP (1 × 104 and 1 × 105) or CKPS (1 × 103 and 1 × 104) cells (n = 4). Data are shown as beeswarm and box plots, and statistical significance was assessed with two-tailed Student t test. Middle, gross appearance of liver metastatic foci. Scale bars, 10 mm. Right, images for hematoxylin and eosin (H&E) staining and PROM1 immunostaining in liver metastatic foci. Scale bars, 100 μm for hematoxylin and eosin staining and 50 μm for PROM1 staining. G, Western blots for verification of Prom1 and Alcam gene knockout by CRISPR in CKPS cells. N.C., negative control for CRISPR. 1 and 2, ID of knockout cell clones. H, Effect of Prom1 or Alcam knockout (KO) on the metastasis-initiating activity of CKPS cells. Left, the number of metastatic foci measured 20 days after injecting 1 × 104 N.C., Prom1 KO, or Alcam KO CKPS cells (n = 6). Middle, representative images of liver metastases. Scale bars, 10 mm. Right, hematoxylin and eosin staining of representative liver metastatic foci. Scale bars, 200 μm. Data are presented as beeswarm and box plots and assessed with one-way ANOVA and Tukey HSD test. I, The effect of Prom1 KO and Alcam KO on 3D-cultured CKPS cells evaluated by spheroid size. Representative micrographs after 15 days of culture (left) and size evaluation (right) after 9 and 15 days of culture. Scale bars, 200 μm for N.C. and 100 μm for Prom1 KO and Alcam KO. *, P < 0.05.

Close modal

We then determined their metastasis-initiating activity in vivo. When transplanted into the spleen and tested for their ability to metastasize to the liver, as few as 1 × 103 CKPS cells were able to initiate metastatic foci in the liver, whereas more than 1 × 105 CKP cells were required to induce substantial metastasis (Fig. 4E and F). Expression of PROM1 was higher in metastatic foci of CKPS cells than those of CKP cells (Fig. 4F), whereas they did not differ in the frequency of proliferating cells or apoptotic cells (Supplementary Fig. S7D). These results demonstrate that CKPS cells express stem cell markers highly and have potent metastasis-initiating activity. To address possible functional involvement of ALCAM and PROM1 in metastasis of intestinal adenocarcinoma cells, we next knocked them out in CKPS cells (Fig. 4G). Alcam KO cells and Prom1 KO cells showed markedly reduced metastasis-initiating activity, whereas there was no significant difference in their proliferation or apoptosis in liver metastatic foci (Fig. 4H; Supplementary Fig. S7E). Prom1 KO and Alcam KO CKPS cells showed significantly lower spheroid-forming activity than control KO cells (Fig. 4I). Intriguingly, most liver metastatic foci induced by transplantation of Alcam KO CKPS cells, as well as those in CKP mice with low ALCAM levels, formed single-layered glandular ducts, in clear contrast to the multilayer glandular ducts observed in metastatic foci induced by CKPS cells or in spontaneous metastases of CKPS mice (Figs. 1D and 4H; Supplementary Fig. S2D). These results indicate the functional importance of ALCAM in liver metastasis of CRC cells and the essential role of SMAD4 in regulating its expression.

The cAMP/PKA/CREB pathway positively regulates expression of ALCAM and PROM1

Our Western blot analysis suggested that additional factors are involved in induction of ALCAM and PROM1 in CKPS spheroid cells (Fig. 4C and D). Therefore, we next examined changes in the phosphorylation status of signaling proteins during spheroid formation. Analysis using a phospho-antibody array showed elevated phosphorylation of CREB, GSK3α, and mTOR, and decreased phosphorylation of GSK3β in CKPS cells under 3D culture, compared with 2D culture (Supplementary Fig. S8A). Western blot analysis confirmed strong induction of CREB phosphorylation in 3D culture (Fig. 5A). Treatment with the CREB inhibitor, 666–15, not only blocked CREB phosphorylation, but also reduced ALCAM level in CKPS spheroids (Fig. 5A), indicating that CREB activation positively regulates ALCAM expression. In contrast, Western blot analysis showed that GSK3β phosphorylation was not reduced under 3D culture (Supplementary Fig. S8B). Treatment with CHR-99021, a GSK inhibitor, elevated wild-type β-catenin, a substrate of GSK, but did not affect levels of ALCAM or PROM1 (Supplementary Fig. S8B). Likewise, we did not observe elevated phosphorylation of S6 in 3D culture (Supplementary Fig. S8C), and treatment with the mTOR kinase inhibitor, AZD8055, failed to affect ALCAM or PROM1 expression (Supplementary Fig. S8C). Thus, we further investigated the role of the cAMP/Protein kinase A (PKA)/CREB pathway.

Figure 5.

The PKA/CREB pathways contribute to plasticity and metastasis-initiating activity of intestinal adenocarcinoma stem cells of CKPS mice. A, Western blots of PROM1, ALCAM, phospho-CREB (p-CREB), p-ATF1, CREB in 2D- and 3D-cultured CKPS cells treated with or without the CREB inhibitor 666–15 for 24 hours. B, Western blots of PROM1 and ALCAM in 2D-cultured CKPS or CKP cells treated with 10 μmol/L forskolin (Fsk) and/or 10 μmol/L SB431542 (SB) for 24 hours. C, Western blots of PROM1, ALCAM, p-CREB, and p-ATF1, and Myc-tagged PKACα in 2D-cultured CKPS cells overexpressing wild-type (WT) or constitutively active PKACα. Mut, constitutively active mutant (H88Q, W197R). D, Western blots of PROM1, ALCAM, p-CREB, and phospho-ATF1 (p-ATF1) in 3D-cultured CKPS cells treated with the adenylyl cyclase inhibitor, MDL-12330A (MDL), the PKA inhibitor H-89 for 24 hours. E, Immunostaining of p-CREB in representative primary tumors (luminal sides and invasion fronts) and liver metastases of CKPS mice. Scale bars, 500 μm for top left and bottom middle panel and 100 μm for other panels. F, Quantification of the percentage of p-CREB-positive glands in primary tumors and liver metastases of CKPS mice. G, Immunostaining of phospho-CREB (p-CREB) in representative clinical samples of the primary tumor (top) and liver metastasis (bottom) from the same patient with colorectal cancer. Scale bars, 500 μm for left panels and 100 μm for right panels showing enlarged views of the frames in the left panels. H, Quantification of the percentage of p-CREB-positive glands in clinical colorectal cancer samples. Middle, the region between luminal-side and invasion front. Data are presented as beeswarm and box plots (F, n = 30 fields from three mice; H, average of three fields of view for each site from 9 patients) and assessed by one-way ANOVA and Tukey HSD test. *, P < 0.05.

Figure 5.

The PKA/CREB pathways contribute to plasticity and metastasis-initiating activity of intestinal adenocarcinoma stem cells of CKPS mice. A, Western blots of PROM1, ALCAM, phospho-CREB (p-CREB), p-ATF1, CREB in 2D- and 3D-cultured CKPS cells treated with or without the CREB inhibitor 666–15 for 24 hours. B, Western blots of PROM1 and ALCAM in 2D-cultured CKPS or CKP cells treated with 10 μmol/L forskolin (Fsk) and/or 10 μmol/L SB431542 (SB) for 24 hours. C, Western blots of PROM1, ALCAM, p-CREB, and p-ATF1, and Myc-tagged PKACα in 2D-cultured CKPS cells overexpressing wild-type (WT) or constitutively active PKACα. Mut, constitutively active mutant (H88Q, W197R). D, Western blots of PROM1, ALCAM, p-CREB, and phospho-ATF1 (p-ATF1) in 3D-cultured CKPS cells treated with the adenylyl cyclase inhibitor, MDL-12330A (MDL), the PKA inhibitor H-89 for 24 hours. E, Immunostaining of p-CREB in representative primary tumors (luminal sides and invasion fronts) and liver metastases of CKPS mice. Scale bars, 500 μm for top left and bottom middle panel and 100 μm for other panels. F, Quantification of the percentage of p-CREB-positive glands in primary tumors and liver metastases of CKPS mice. G, Immunostaining of phospho-CREB (p-CREB) in representative clinical samples of the primary tumor (top) and liver metastasis (bottom) from the same patient with colorectal cancer. Scale bars, 500 μm for left panels and 100 μm for right panels showing enlarged views of the frames in the left panels. H, Quantification of the percentage of p-CREB-positive glands in clinical colorectal cancer samples. Middle, the region between luminal-side and invasion front. Data are presented as beeswarm and box plots (F, n = 30 fields from three mice; H, average of three fields of view for each site from 9 patients) and assessed by one-way ANOVA and Tukey HSD test. *, P < 0.05.

Close modal

To determine the effect of cAMP elevation on stem cell properties of colorectal cancer cells, we then treated CKPS cells in 2D culture with forskolin (Fsk), an adenylyl cyclase (ADCY) activator. Fsk treatment robustly increased both PROM1 and ALCAM levels in CKPS cells (Fig. 5B). On the other hand, although a single treatment with Fsk only slightly increased proteins levels of PROM1 and ALCAM in CKP cells, the combination treatment of Fsk and the TGFβ receptor inhibitor, SB431542, synergistically increased their levels (Fig. 5B). Overexpression of the catalytic subunit α of PKA (PKA Cα) induced Alcam expression, and the constitutively active mutant of PKA Cα (H88Q, W197R) increased expression of not only ALCAM, but also PROM1 (Fig. 5C). Although it is unclear why the constitutively active mutant of PKA Cα induced lower CREB phosphorylation than wild-type PKA Cα, we speculate that a feedback mechanism or another pathway that allows accumulation of PROM1 protein may be involved. Treatment with the ADCY inhibitor, MDL-12330A, or the PKA inhibitor, H-89, suppressed ALCAM levels in spheroids (Fig. 5D). IHC analysis showed that the frequency of phospho-CREB-positive tumor epithelial cells was much higher in the primary tumor invasion front and liver metastases compared with the luminal side of the primary tumors (Fig. 5E and F). Immunostaining of clinical samples of primary tumors and liver metastases from the same patients with colorectal cancer also showed phospho-CREB-positive cancer epithelial cells in the invasion fronts of primary tumors and metastatic lesions (Fig. 5G and H). Taken together, these data demonstrate that expression of ALCAM and PROM1 is regulated positively and negatively by the cAMP/PKA/CREB and TGFβ/SMAD4 pathways, respectively, in intestinal adenocarcinoma cells of CKPS mice.

CREB inhibition reduces the metastatic potential of colorectal cancer cells and enhances irinotecan-induced suppression of colorectal cancer metastasis

To determine the function of CREB in maintaining plasticity and metastatic activity, we then generated Creb KO CKPS cells. When Creb KO CKPS cells were transplanted into the spleen, the number of liver metastatic foci was markedly reduced compared with control cells (Fig. 6A). Consistently, treatment with the CREB inhibitor, 666–15, also significantly suppressed the metastatic ability of CKPS cells (Fig. 6B). We next explored whether co-treatment with 666–15 could affect efficacy of chemotherapy for colorectal cancer metastasis. A single treatment with irinotecan, a topoisomerase inhibitor used in treatment of colorectal cancer, significantly, but weakly reduced the number of metastatic foci (Fig. 6C). Interestingly, liver metastases of irinotecan-treated mice showed increased phosphorylation of CREB and expression of PROM1 compared with those of untreated mice (Fig. 6D; Supplementary Figs. S9A and S9B). Combination treatment of irinotecan and 666–15 was significantly more effective in inhibiting liver metastasis of CKPS cells than irinotecan alone, accompanied by reduction of CREB phosphorylation and PROM1 expression (Fig. 6E; Supplementary Figs. S9C and S9D). Taken together, these data demonstrate that expression of ALCAM is regulated positively and negatively by the cAMP/PKA/CREB and TGFβ/SMAD4 pathways, respectively, in intestinal adenocarcinoma cells of CKPS mice and suggest that the former pathway may be a potential therapeutic target for metastatic colorectal cancer (Fig. 6F).

Figure 6.

CREB inhibition suppresses liver metastasis of colorectal cancer cells. A, Effect of Creb knockout on the metastasis-initiating activity of CKPS cells. Left top, Western blots of PROM1 and ALCAM in Creb KO CKPS cells. Left bottom, representative images of liver metastases. Scale bars, 5 mm. Right, the number of metastatic foci measured 20 days after injecting 1 × 104 of N.C. or Creb KO CKPS cells (n = 6). N.C., negative control for CRISPR. B, Effect of the CREB inhibitor, 666–15, on the metastasis-initiating activity of CKPS cells. Left top, schematic diagram of the 666–15 treatment schedule. Left bottom, representative images of liver metastases. Scale bars, 5 mm. Right, the number of metastatic foci measured 20 days after injecting CKPS cells with or without 666–15 treatment (n = 6). C, Effect of irinotecan on the metastasis-initiating activity of CKPS cells. Left top, schematic diagram of irinotecan treatment schedule. Left bottom, representative images of liver metastases. Scale bars, 5 mm. Right, the number of metastatic foci measured 28 days after injecting CKPS cells treated with or without irinotecan (n = 6). D, Western blots of phospho-CREB (p-CREB), p-ATF1 in liver metastasis of CKPS cells from C. E, Effect of combined treatment of irinotecan with 666–15 on the metastasis-initiating activity of CKPS cells. Left top, schematic diagram of irinotecan treatment schedule. Left bottom, representative images of liver metastases. Scale bars, 5 mm. Right, the number of metastatic foci measured 28 days after injecting CKPS cells treated with irinotecan alone (n = 6) or irinotecan plus 666–15 (n = 8). Data are presented as beeswarm and box plots (A, B, C and E), and assessed with two-tailed Student t test. F, Schematic representation of pathways involved in the stemness of colorectal cancer cells. ADCY, adenylate cyclase. *, P < 0.05.

Figure 6.

CREB inhibition suppresses liver metastasis of colorectal cancer cells. A, Effect of Creb knockout on the metastasis-initiating activity of CKPS cells. Left top, Western blots of PROM1 and ALCAM in Creb KO CKPS cells. Left bottom, representative images of liver metastases. Scale bars, 5 mm. Right, the number of metastatic foci measured 20 days after injecting 1 × 104 of N.C. or Creb KO CKPS cells (n = 6). N.C., negative control for CRISPR. B, Effect of the CREB inhibitor, 666–15, on the metastasis-initiating activity of CKPS cells. Left top, schematic diagram of the 666–15 treatment schedule. Left bottom, representative images of liver metastases. Scale bars, 5 mm. Right, the number of metastatic foci measured 20 days after injecting CKPS cells with or without 666–15 treatment (n = 6). C, Effect of irinotecan on the metastasis-initiating activity of CKPS cells. Left top, schematic diagram of irinotecan treatment schedule. Left bottom, representative images of liver metastases. Scale bars, 5 mm. Right, the number of metastatic foci measured 28 days after injecting CKPS cells treated with or without irinotecan (n = 6). D, Western blots of phospho-CREB (p-CREB), p-ATF1 in liver metastasis of CKPS cells from C. E, Effect of combined treatment of irinotecan with 666–15 on the metastasis-initiating activity of CKPS cells. Left top, schematic diagram of irinotecan treatment schedule. Left bottom, representative images of liver metastases. Scale bars, 5 mm. Right, the number of metastatic foci measured 28 days after injecting CKPS cells treated with irinotecan alone (n = 6) or irinotecan plus 666–15 (n = 8). Data are presented as beeswarm and box plots (A, B, C and E), and assessed with two-tailed Student t test. F, Schematic representation of pathways involved in the stemness of colorectal cancer cells. ADCY, adenylate cyclase. *, P < 0.05.

Close modal

In this study, we created a novel, genetically-engineered mouse model (GEMM) of colorectal cancer that spontaneously develops liver metastasis. Various mouse models of colorectal cancer liver metastasis have been developed previously, including transplantation models using cell lines, organoids, or patient-derived xenografts, as well as GEMMs (6, 29). Relatively few intestinal tumors in our model, induced by spontaneous activation of CreERT2, prolonged survival long enough to develop metastases, eliminating the need for artificial interventions such as overexpression of mutant Kras, or induction of colitis by colonic injection of trypsin or by DSS treatment as employed in previous GEMMs of metastatic colorectal cancer (30–32). Thus, CKPS mice are likely to better reflect the natural course of colorectal cancer progression, including stromal responses in both primary and metastatic lesions, and may serve as a practical, workable model for in-depth analysis of metastatic mechanisms and for evaluating treatment options.

No additional DNA mutations implicated in colorectal cancer progression were found in liver metastases compared with primary tumors of CKPS mice, consistent with previous reports (4, 32–35). On the other hand, we found increased expression of known colorectal cancer stem cell markers in primary tumors of CKPS mice, compared with nonmetastatic tumors. Metastasis is thought to be initiated by cancer cells with stem-like properties (5, 35, 36). Interestingly, our immunohistochemical analyses showed that expression of CDX2, an intestinal differentiation marker, was reduced, whereas levels of SOX17, a definitive endoderm marker, ALCAM, a colorectal cancer stem-cell marker, and phospho-CREB were elevated in invasion fronts, compared with luminal sides of primary tumors. These results suggest that the microenvironment in invasion fronts may confer stem-like properties essential for metastasis of colorectal cancer cells (Fig. 6F).

Meta-analyses of colorectal cancer demonstrated that higher levels of PROM1 are associated with poor clinical outcomes, and high ALCAM levels are associated with poor prognoses and distant metastases (37, 38); however, their functional involvement in colorectal cancer progression and metastasis remains obscure. Knockout of PROM1 or ALCAM in CKPS cells suppressed their spheroid-formation in vitro and metastasis formation in vivo, indicating its essential role in plasticity and metastasis initiation activity of colorectal cancer cells. Because ALCAM acts as an adhesion molecule between cells, and liver metastases formed by Alcam KO CKPS cells showed single-layered glandular histology (Fig. 4H), ALCAM may promote expansion of CKPS cells in the liver by enhancing cell–cell adhesion (39). With regard to involvement of cell adhesion molecules in stemness and metastasis of colorectal cancer cells, SOSTDC1 promoted metastasis of colorectal cancer cells via interaction with ALCAM (40), and the L1 cell adhesion molecule (L1CAM) proved essential in metastasis-initiating cells of colorectal cancer (41). However, our RNA-seq data showed low levels of Sostdc1 and L1cam expression in primary and metastatic tumors of CKPS mice, and we did not pursue their involvement in this model. Knockout of PROM1 also reduced the metastatic potential of CKPS cells. Although PROM1 reportedly enhances migration of SW620 colorectal cancer cells via SRC (42), its functional role in metastasis is unclear. Future studies using our model may help to illuminate molecular mechanisms by which ALCAM and PROM1 contribute to colorectal cancer metastasis.

We observed a much higher frequency of liver metastasis in CKPS mice compared with CKP mice. Our data showed that CKPS cells had more potent metastasis-initiating activity than CKP cells and that the TGFβ/SMAD4 pathway negatively regulates PROM1 and ALCAM, suggesting that SMAD4 loss contributes to metastasis by promoting plasticity of colorectal cancer stem cells. However, SMAD4 loss is not sufficient for metastasis, as indicated by the absence of liver metastasis in cis-Apc/Smad4 (AS) mice (13), CKS mice, or CPS mice (Fig. 1E). Because intestinal adenocarcinoma cells in AS mice contain many differentiated tumor cells that produce acidic mucus (13), we speculate that mutations in p53 and Kras mutations are essential for survival of colorectal cancer cells with preserved stemness. The TGFβ/SMAD4 pathway is involved in regulating differentiation of embryonic or pluripotent stem cells into IECs, and inhibition of the pathway reprograms differentiated cells into pluripotent stem cells and endoderm cells (43, 44). Our study also revealed key roles of the cAMP/PKA/CREB pathway in ALCAM expression, as well as in regulating stemness and metastatic potential of colorectal cancer cells. This pathway has been implicated in hematopoietic/leukemic stem cells (45), but its involvement in solid cancer stem cells has remained uncertain. Taken together, our present results using a novel metastasis model of colorectal cancer demonstrate vital roles of the cAMP/PKA/CREB and TGFβ/SMAD4 pathways in maintaining plasticity and metastasis-initiating activity of colorectal cancer stem cells via ALCAM and PROM1. Importantly, co-treatment with a CREB inhibitor and irinotecan, a cytotoxic agent used for colorectal cancer chemotherapy, strongly suppressed liver metastasis of CKPS cells. Further study of this pathway in colorectal cancer stem cells may lead to novel therapeutic strategies for metastatic colorectal cancer, as well as its relapse and drug resistance.

T. Fujishita reports grants from Japan Society for the Promotion of Science (JSPS), Nagono Medical Foundation, MSD Life Science Foundation, and Daiwa Securities Health Foundation during the conduct of the study. M. Aoki reports grants from JSPS, Japan Agency for Medical Research and Development, and Takeda Science Foundation during the conduct of the study. No disclosures were reported by the other authors.

T. Fujishita: Conceptualization, resources, data curation, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. Y. Kojima: Conceptualization, data curation, formal analysis, funding acquisition, investigation, visualization, writing–original draft, project administration, writing–review and editing. R. Kajino-Sakamoto: Data curation, formal analysis, investigation, visualization. E. Mishiro-Sato: Data curation, investigation. Y. Shimizu: Resources, data curation, investigation. W. Hosoda: Resources. R. Yamaguchi: Resources, investigation. M.M. Taketo: Resources, investigation. M. Aoki: Conceptualization, resources, supervision, funding acquisition, investigation, project administration, writing–review and editing.

The authors thank Kyoko Kobori, Yoshiko Goto, Miyako Tsuda, and Yu Li for technical assistance, and Kentaro Taki (Division for Medical Research Engineering, Nagoya University Graduate School of Medicine) for technical support of LC/MS. This work was supported by JSPS KAKENHI Grant Numbers JP20K07440 (to T. Fujishita), JP18H02686 (to M. Aoki), 22H02909 (to M. Aoki), the Project for Cancer Research and Therapeutics Evolution (P-CREATE) from the Japan Agency for Medical Research and Development (to M. Aoki, 21cm0106282h0001), and by grants from the Nagono Medical Foundation (to T. Fujishita), the MSD Life Science Foundation (to T. Fujishita), the Daiwa Securities Health Foundation (to T. Fujishita), and the Takeda Science Foundation (to M. Aoki).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

1.
Bray
F
,
Ferlay
J
,
Soerjomataram
I
,
Siegel
RL
,
Torre
LA
,
Jemal
A
.
Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries
.
CA Cancer J Clin
2018
;
68
:
394
424
.
2.
Siegel
RL
,
Miller
KD
,
Fuchs
HE
,
Jemal
A
.
Cancer Statistics, 2021
.
CA Cancer J Clin
2021
;
71
:
7
33
.
3.
Di Nicolantonio
F
,
Vitiello
PP
,
Marsoni
S
,
Siena
S
,
Tabernero
J
,
Trusolino
L
, et al
.
Precision oncology in metastatic colorectal cancer: from biology to medicine
.
Nat Rev Clin Oncol
2021
;
18
:
506
25
.
4.
Yaeger
R
,
Chatila
WK
,
Lipsyc
MD
,
Hechtman
JF
,
Cercek
A
,
Sanchez-Vega
F
, et al
.
Clinical sequencing defines the genomic landscape of metastatic colorectal cancer
.
Cancer Cell
2018
;
33
:
125
36
.
5.
Ganesh
K
,
Massagué
J
.
Targeting metastatic cancer
.
Nat Med
2021
;
27
:
34
44
.
6.
Andres
SF
,
Williams
KN
,
Rustgi
AK
.
The molecular basis of metastatic colorectal cancer
.
Curr Colorectal Cancer Rep
2018
;
14
:
69
79
.
7.
Bakir
B
,
Chiarella
AM
,
Pitarresi
JR
,
Rustgi
AK
.
EMT, MET, plasticity, and tumor metastasis
.
Trends Cell Biol
2020
;
30
:
764
76
.
8.
Medema
JP
.
Targeting the colorectal cancer stem cell
.
N Engl J Med
2017
;
377
:
888
90
.
9.
Brodt
P
.
Role of the microenvironment in liver metastasis: from pre- to prometastatic niches
.
Clin Cancer Res
2016
;
22
:
5971
82
.
10.
Celià-Terrassa
T
,
Kang
Y
.
Metastatic niche functions and therapeutic opportunities
.
Nat Cell Biol
2018
;
20
:
868
77
.
11.
Kristianto
J
,
Johnson
MG
,
Zastrow
RK
,
Radcliff
AB
,
Blank
RD
.
Spontaneous recombinase activity of Cre-ERT2 in vivo
.
Transgenic Res
2017
;
26
:
411
7
.
12.
Harada
N
,
Tamai
Y
,
Ishikawa
T
,
Sauer
B
,
Takaku
K
,
Oshima
M
, et al
.
Intestinal polyposis in mice with a dominant stable mutation of the beta-catenin gene
.
EMBO J
1999
;
18
:
5931
42
.
13.
Takaku
K
,
Oshima
M
,
Miyoshi
H
,
Matsui
M
,
Seldin
MF
,
Taketo
MM
.
Intestinal tumorigenesis in compound mutant mice of both Dpc4 (Smad4) and Apc genes
.
Cell
1998
;
92
:
645
56
.
14.
el Marjou
F
,
Janssen
K-P
,
Chang
BH-J
,
Li
M
,
Hindie
V
,
Chan
L
, et al
.
Tissue-specific and inducible Cre-mediated recombination in the gut epithelium
.
Genesis
2004
;
39
:
186
93
.
15.
Schneider
CA
,
Rasband
WS
,
Eliceiri
KW
.
NIH Image to ImageJ: 25 years of image analysis
.
Nat Methods
2012
;
9
:
671
5
.
16.
Fujishita
T
,
Aoki
M
,
Taketo
MM
.
JNK signaling promotes intestinal tumorigenesis through activation of mTOR complex 1 in Apc(Δ716) mice
.
Gastroenterology
2011
;
140
:
1556
63
.e6.
17.
Fujishita
T
,
Kojima
Y
,
Kajino-Sakamoto
R
,
Taketo
MM
,
Aoki
M
.
Tumor microenvironment confers mTOR inhibitor resistance in invasive intestinal adenocarcinoma
.
Oncogene
2017
;
36
:
6480
9
.
18.
R Core Team
.
R: A language and environment for statistical computing
.
Vienna, Austria
:
R Foundation for Statistical Computing
;
2020
.
Available from
: https://www.R-project.org/.
19.
CGA Network
.
Comprehensive molecular characterization of human colon and rectal cancer
.
Nature
2012
;
487
:
330
7
.
20.
Coskun
M
,
Troelsen
JT
,
Nielsen
OH
.
The role of CDX2 in intestinal homeostasis and inflammation
.
Biochim Biophys Acta
2011
;
1812
:
283
9
.
21.
Dalerba
P
,
Sahoo
D
,
Paik
S
,
Guo
X
,
Yothers
G
,
Song
N
, et al
.
CDX2 as a prognostic biomarker in stage II and stage III colon cancer
.
N Engl J Med.
2016
;
374
:
211
22
.
22.
Muñoz
J
,
Stange
DE
,
Schepers
AG
,
van de Wetering
M
,
Koo
B-K
,
Itzkovitz
S
, et al
.
The Lgr5 intestinal stem cell signature: robust expression of proposed quiescent “+4” cell markers
.
EMBO J
2012
;
31
:
3079
91
.
23.
Dalerba
P
,
Dylla
SJ
,
Park
I-K
,
Liu
R
,
Wang
X
,
Cho
RW
, et al
.
Phenotypic characterization of human colorectal cancer stem cells
.
Proc Natl Acad Sci U S A
2007
;
104
:
10158
63
.
24.
Vermeulen
L
,
Todaro
M
,
de Sousa Mello
F
,
Sprick
MR
,
Kemper
K
,
Perez Alea
M
, et al
.
Single-cell cloning of colon cancer stem cells reveals a multi-lineage differentiation capacity
.
Proc Natl Acad Sci U S A
2008
;
105
:
13427
32
.
25.
Sasaki
N
,
Sachs
N
,
Wiebrands
K
,
Ellenbroek
SIJ
,
Fumagalli
A
,
Lyubimova
A
, et al
.
Reg4+ deep crypt secretory cells function as epithelial niche for Lgr5+ stem cells in colon
.
Proc Natl Acad Sci U S A
2016
;
113
:
E5399
407
.
26.
Schewe
M
,
Franken
PF
,
Sacchetti
A
,
Schmitt
M
,
Joosten
R
,
Böttcher
R
, et al
.
Secreted phospholipases A2 are intestinal stem cell niche factors with distinct roles in homeostasis, inflammation, and cancer
.
Cell Stem Cell
2016
;
19
:
38
51
.
27.
Barker
N
,
van Es
JH
,
Kuipers
J
,
Kujala
P
,
van den Born
M
,
Cozijnsen
M
, et al
.
Identification of stem cells in small intestine and colon by marker gene Lgr5
.
Nature
2007
;
449
:
1003
7
.
28.
Weiswald
L-B
,
Bellet
D
,
Dangles-Marie
V
.
Spherical cancer models in tumor biology
.
Neoplasia
2015
;
17
:
1
15
.
29.
Oh
BY
,
Hong
HK
,
Lee
WY
,
Cho
YB
.
Animal models of colorectal cancer with liver metastasis
.
Cancer Lett
2017
;
387
:
114
20
.
30.
Boutin
AT
,
Liao
W-T
,
Wang
M
,
Hwang
SS
,
Karpinets
TV
,
Cheung
H
, et al
.
Oncogenic Kras drives invasion and maintains metastases in colorectal cancer
.
Genes Dev
2017
;
31
:
370
82
.
31.
Hung
KE
,
Maricevich
MA
,
Richard
LG
,
Chen
WY
,
Richardson
MP
,
Kunin
A
, et al
.
Development of a mouse model for sporadic and metastatic colon tumors and its use in assessing drug treatment
.
Proc Natl Acad Sci U S A
2010
;
107
:
1565
70
.
32.
Tauriello
DVF
,
Palomo-Ponce
S
,
Stork
D
,
Berenguer-Llergo
A
,
Badia-Ramentol
J
,
Iglesias
M
, et al
.
TGFβ drives immune evasion in genetically reconstituted colon cancer metastasis
.
Nature
2018
;
554
:
538
43
.
33.
Sakai
E
,
Nakayama
M
,
Oshima
H
,
Kouyama
Y
,
Niida
A
,
Fujii
S
, et al
.
Combined mutation of Apc, Kras, and Tgfbr2 effectively drives metastasis of intestinal cancer
.
Cancer Res.
2018
;
78
:
1334
46
.
34.
Fumagalli
A
,
Drost
J
,
Suijkerbuijk
SJE
,
van Boxtel
R
,
de Ligt
J
,
Offerhaus
GJ
, et al
.
Genetic dissection of colorectal cancer progression by orthotopic transplantation of engineered cancer organoids
.
Proc Natl Acad Sci U S A
2017
;
114
:
E2357
64
.
35.
Massagué
J
,
Ganesh
K
.
Metastasis-initiating cells and ecosystems
.
Cancer Discov
2021
;
11
:
971
94
.
36.
Fares
J
,
Fares
MY
,
Khachfe
HH
,
Salhab
HA
,
Fares
Y
.
Molecular principles of metastasis: a hallmark of cancer revisited
.
Signal Transduct Target Ther
2020
;
5
:
28
.
37.
Chen
S
,
Song
X
,
Chen
Z
,
Li
X
,
Li
M
,
Liu
H
, et al
.
CD133 expression and the prognosis of colorectal cancer: a systematic review and meta-analysis
.
PLoS One
2013
;
8
:
e56380
.
38.
Zhang
Y
,
Qian
C
,
Jing
L
,
Ren
J
,
Guan
Y
.
Meta-analysis indicating that high ALCAM expression predicts poor prognosis in colorectal cancer
.
Oncotarget
2017
;
8
:
48272
81
.
39.
Fernández
MM
,
Ferragut
F
,
Cárdenas Delgado
VM
,
Bracalente
C
,
Bravo
AI
,
Cagnoni
AJ
, et al
.
Glycosylation-dependent binding of galectin-8 to activated leukocyte cell adhesion molecule (ALCAM/CD166) promotes its surface segregation on breast cancer cells
.
Biochim Biophys Acta
2016
;
1860
:
2255
68
.
40.
Bartolomé
RA
,
Pintado-Berninches
L
,
Jaén
M
,
de Los Ríos
V
,
Imbaud
JI
,
Casal
JI
.
SOSTDC1 promotes invasion and liver metastasis in colorectal cancer via interaction with ALCAM/CD166
.
Oncogene
2020
;
39
:
6085
98
.
41.
Ganesh
K
,
Basnet
H
,
Kaygusuz
Y
,
Laughney
AM
,
He
L
,
Sharma
R
, et al
.
L1CAM defines the regenerative origin of metastasis-initiating cells in colorectal cancer
.
Nat Cancer
2020
;
1
:
28
45
.
42.
Liu
C
,
Li
Y
,
Xing
Y
,
Cao
B
,
Yang
F
,
Yang
T
, et al
.
The Interaction between cancer stem cell marker CD133 and Src protein promotes focal adhesion kinase (FAK) phosphorylation and cell migration
.
J Biol Chem
2016
;
291
:
15540
50
.
43.
Cao
S
,
Yu
S
,
Chen
Y
,
Wang
X
,
Zhou
C
,
Liu
Y
, et al
.
Chemical reprogramming of mouse embryonic and adult fibroblast into endoderm lineage
.
J Biol Chem
2017
;
292
:
19122
32
.
44.
Spence
JR
,
Mayhew
CN
,
Rankin
SA
,
Kuhar
MF
,
Vallance
JE
,
Tolle
K
, et al
.
Directed differentiation of human pluripotent stem cells into intestinal tissue in vitro
.
Nature
2011
;
470
:
105
9
.
45.
Cheng
JC
,
Kinjo
K
,
Judelson
DR
,
Chang
J
,
Wu
WS
,
Schmid
I
, et al
.
CREB is a critical regulator of normal hematopoiesis and leukemogenesis
.
Blood
2008
;
111
:
1182
92
.