Cancer stem cells are implicated in tumor progression, metastasis, and recurrence, although the exact mechanisms remain poorly understood. Here, we show that the expression of cellular prion protein (PrPc, PRNP) is positively correlated with an increased risk of metastasis in colorectal cancer. PrPc defines a subpopulation of CD44-positive cancer stem cells that contributes to metastatic capacity. PrPc+CD44+ colorectal cancer stem cells displayed high liver metastatic capability, unlike PrPcCD44+ stem cells, that was inhibited by RNAi-mediated attenuation of PrPc. Notably, administration of PrPc monoclonal antibodies significantly inhibited tumorigenicity and metastasis of colorectal cancer stem cells in mouse models of orthotopic metastasis. PrPc promoted epithelial to mesenchymal transition (EMT) via the ERK2 (MAPK1) pathway, thereby conferring high metastatic capacity. Our findings reveal the function of PrPc in regulating EMT in cancer stem cells, and they identify PrPc as candidate therapeutic target in metastatic colorectal cancer. Cancer Res; 73(8); 2682–94. ©2013 AACR.

Cancer is a major cause of mortality, and most human cancer-related deaths are related to metastasis (1, 2). The difficulty in diagnosing and therapeutically eliminating metastasis is related to our limited understanding of the molecules specifically expressed on the surface of metastatic cancer cells. Accumulating evidence suggests that a subset of cancer cells, operationally termed cancer stem cells (CSC) or tumor-initiating cells, possesses enhanced ability to regenerate tumors and survive anticancer therapies (3). CSCs were first identified in the hematopoietic system (4), and have now been reported in most human solid tumors on the basis of their functional properties (5–7). Colorectal CSCs (CCSC) have been distinguished using CD44 or CD133 either alone (8–10), or combined with other markers, such as EpCAM, CD166, CD29, CD24, LGR5 and aldehyde dehydrogenase 1 (ALDH1; refs. 11–13). Previous studies by ourselves and other authors have shown that as few as 100 CD44+ cancer cells isolated from patients were able to develop into a heterogeneous tumor, and that spheroids derived from a single CD44+ cancer cell could recapitulate the heterogeneous hierarchy of tumor cells (8, 14). These studies indicate that CCSCs may trigger the formation of colorectal cancer (CRC), and that CD44 is a functionally important surface biomarker of CCSCs. Importantly, an antibody specific to CD44 selectively eradicated leukemic stem cells in vivo by altering the fate of leukemic stem cells (15). Monoclonal antibodies targeting surface antigens specifically expressed in CSCs offer a potentially powerful strategy to eliminate CSCs with minimal toxicity. Besides CD44, monoclonal antibodies targeting CD123 and CD47 have been reported to eradicate leukemia stem cells (16).

Recent evidence suggests that CSCs may be endowed with enhanced disseminating capacity enabling them to invade neighboring tissues or metastasize to distant organs (17–19). Although CD44 and some other adhesion molecules, such as E-cadherin, could act as migration suppressors by restricting cancer cells to their primary sites (20), these same molecules with certain modifications (e.g., splice variants of CD44) or in combination with other adhesion molecules, such as CD26, may result in metastatic phenotypes (21–24).

Cellular prion protein (PrPc) is a highly conserved glycoprotein present in all vertebrates and has the same protein sequence as the scrapie prion protein (PrPsc), a pathogenic factor in scrapie in sheep, bovine spongiform encephalopathy in cattle, and kuru in human beings (25, 26). In addition to the central involvement of PrPc in prion diseases, several lines of evidence have intriguingly implicated PrPc in human tumors including glioblastoma, breast, prostate, gastric and colorectal cancers (CRC; 27). PrPc interacts homophilically with itself or heterotypically with components of the extracellular matrix (ECM) such as laminin and vitronectin, to mediate cell adhesion (28). PrPc expression is required for the self-renewal of CD34+ hematopoietic stem cells (29, 30). However, despite tantalizing evidence linking PrPc to human cancers, its physiologic function and exact role in cancer development remain elusive. Here, we report that high levels of PrPc expression (in PrPc+ cells) identify a functionally distinct subpopulation of CD44+ CCSCs, which contributes to CRC initiation and metastasis. Our findings thus reveal hitherto-unknown roles of PrPc in CCSCs and epithelial–mesenchymal transition (EMT), and suggest that PrPc could provide a potential therapeutic target in metastatic CRCs.

Cells, animals, and samples

Fresh colon samples from patients with CRC and paired normal subjects were collected from the tumor bank of Beijing Cancer Hospital (Beijing, China), as approved by the Research Ethics Board at the Beijing Institute for Cancer Research. A tissue microarray (TMA) was constructed as described previously (31, 32). The NCI-60 human tumor cell line panel was kindly provided by Dr. Jacque Grassi (National Cancer Institute, Bethesda, MD). SW480 and MDA-MB-231 cell lines were purchased from American Type Culture Collection, authenticated using polyphasic (genotypic and phenotypic) testing every 6 months, and cultured according to the manufacturer's instructions. Human primary CRC cells were isolated from patients and cultured in with serum-free medium. All CCSCs were subcultured for fewer than 20 passages. Four-week-old female nude mice (BALB/c-nu/nu) and 5-week-old female Nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice were purchased from Chinese Academy of Medical Sciences (Beijing, China), and all experiments were conducted under standard conditions in accordance with the National and Institutional regulations.

Tumorigenesis and metastasis in orthotopic xenograft models

NOD/SCID mice were anesthetized with pentobarbital and the cecums were exteriorized by laparotomy. Various numbers of CD44+PrPc+, CD44+PrPc, CD44PrPc+, and CD44PrPc subpopulations freshly sorted from patients, or 50,000 DsRed-labeled PrPc overexpressing or knockdown CCSCs were injected into the cecal wall. All the animals were monitored for up to 80 to 100 days or until tumor-associated death occurred. Primary tumors and metastasis were evaluated by a whole-body fluorescence imaging system (In-Vivo FX PRO, Carestream).

The PrPc expression level is correlated with clinical and pathologic features and outcomes in CRC patients

On the basis of our previous observations that CD44 is a robust marker for CCSCs (14), we hypothesized that additional factors may exist that are coexpressed with CD44 in the promotion of migration and metastasis of CSCs. We therefore compared gene expression in colorectal cancer, liver metastases, and normal tissues by cDNA array conducted in our lab. In addition, we conducted a CD44 coexpression analysis from Oncomine datasets (Ki_Colon, Koinuma_Colon, and Smith_ Colorectal_2 datasets; refs., 33–35). Interestingly, we found that several genes were in fact coexpressed with CD44 and upregulated in metastases; one of these genes was PRNP (data not shown), which encodes PrPc.

To find out whether PrPc and CD44 are associated with the clinical and pathologic features, we carried out immunohistochemical (IHC) staining in a TMA containing 309 colorectal tumor specimens. CD44 and PrPc were found to be coexpressed in most of the tumors, especially in those that were poorly differentiated (correlation coefficient = 0.73; Fig. 1A). The PrPc expression level was significantly higher in poorly differentiated than in moderately and well-differentiated samples (P < 0.001; Fig. 1A and Supplementary Fig. S1A), whereas the level of CD44 was not associated with clinical stage (P = 0.143; Fig. 1A). This observation was supported by examining multiple Oncomine cDNA microarray datasets, which revealed that the PRNP mRNA expression levels were significantly elevated in high-grade and advanced-stage CRCs (Supplementary Fig. S1B). Importantly, the patients with CRC with the highest PrPc expression were also found to have the poorest prognosis, with a median survival of only 12 months (Fig. 1B), which was significantly (P < 0.0001) shorter than that of patients whose primary tumor was PrPc-negative (median survival >5 years).

Figure 1.

PrPc is associated with high grade tumors and poor prognosis in patients. A, heatmap of CD44 and PrPc protein expression in CRCs (n = 309 patients). Data are grouped by the following tumor grades: grade 1, well-differentiated; grade 2, moderately differentiated; grade 3, poorly differentiated. CD44 and PrPc expression are semiquantitatively described as 0, negative; 1, low; 2, moderate; and 3, high. B, Kaplan–Meier plot showing the survival percentage of 109 patients with CRC (>5 years after surgery) grouped by PrPc expression level. The number of patients (n) in each group is indicated in parentheses. C, the PrPc expression level is positively correlated with metastasis. Patients with a concurrent metastasis (M, red) and those without a metastasis (Non-M, blue) are grouped by PrPc expression levels. y-axis, number of patients. **, P < 0.001. D, double immunofluorescence staining of cytokeratin 20 (green) and PrPc (red) in CRC showing preferential localization of PrPc-positive cells in the invasive fronts of tumor islets (indicated by dotted white lines). Scale bar, 200 μm. E, representative immunofluorescence images of PrPc expression in invasive vasculatures in the draining lymph node (white dotted outline). PrPc+/cytokeratin 20+ tumor cells (yellow) are magnified and indicated (arrows). Scale bar, 200 μm. F, IHC staining shows PrPc+ CRC cell invasion into stromal cells (red arrows). Scale bar, 100 μm.

Figure 1.

PrPc is associated with high grade tumors and poor prognosis in patients. A, heatmap of CD44 and PrPc protein expression in CRCs (n = 309 patients). Data are grouped by the following tumor grades: grade 1, well-differentiated; grade 2, moderately differentiated; grade 3, poorly differentiated. CD44 and PrPc expression are semiquantitatively described as 0, negative; 1, low; 2, moderate; and 3, high. B, Kaplan–Meier plot showing the survival percentage of 109 patients with CRC (>5 years after surgery) grouped by PrPc expression level. The number of patients (n) in each group is indicated in parentheses. C, the PrPc expression level is positively correlated with metastasis. Patients with a concurrent metastasis (M, red) and those without a metastasis (Non-M, blue) are grouped by PrPc expression levels. y-axis, number of patients. **, P < 0.001. D, double immunofluorescence staining of cytokeratin 20 (green) and PrPc (red) in CRC showing preferential localization of PrPc-positive cells in the invasive fronts of tumor islets (indicated by dotted white lines). Scale bar, 200 μm. E, representative immunofluorescence images of PrPc expression in invasive vasculatures in the draining lymph node (white dotted outline). PrPc+/cytokeratin 20+ tumor cells (yellow) are magnified and indicated (arrows). Scale bar, 200 μm. F, IHC staining shows PrPc+ CRC cell invasion into stromal cells (red arrows). Scale bar, 100 μm.

Close modal

CRC-related death is usually attributable to recurrence and distant metastasis. Out of 309 specimens, there were 228 primary CRCs without metastasis versus 81 cases with concurrent distant metastasis. Strikingly, we found that PrPc was highly expressed in the primary tumors of the patients with liver or lung metastasis (P = 0.0007; Fig. 1C). Oncomine cDNA microarray data analysis also showed a close correlation of the PrPc level with disease recurrence and early patient death (Supplementary Fig. S1C). We also observed that PrPc expression was most prominent at the interface and invasion fronts of tumor nodules (identified by cytokeratin 20 staining; Fig. 1D), which may constitute a reservoir of mobile CSCs (36). In the lymph nodes that had been invaded by CRC cells, PrPc+ cells were also observed in vascular invasions (Fig. 1E and Supplementary Fig. S1D). In addition, high levels of PrPc were present in CRC cells and could be seen invading the tumor stroma (Fig. 1F). Collectively, these data indicate that high PrPc expression is associated with advanced clinical stages, increased metastatic risk, and reduced patient survival.

A subpopulation of PrPc+ CRC cells possesses CSC properties

We next examined PrPc and CD44 coexpression patterns at the cellular level, and determined the biologic and tumorigenic properties of PrPc-expressing cells in CRC. We detected CD44+PrPc+ cells in the cryosections of CRCs by double immunofluorescence (IF) staining of CD44 and PrPc (Fig. 2A). Flow cytometry analysis revealed only a small percentage of PrPc+ cells in the CD45EpCAM+CD44+ population in patient tumors (Fig. 2B), and the abundance of CD44+PrPc+ cells ranged from 0.02% to 10.83% (median 0.28%; Fig. 2C and Supplementary Table S1). Interestingly, the subpopulation of CD44+PrPc+ cells in the 6 patients with concomitant liver metastasis was significantly higher than that in the 25 primary CRC patients without metastasis (3.19% vs. 0.79%, mean, P = 0.03; Supplementary Table S1). Furthermore, PrPc and CD44 were closely associated in the NCI-60 human tumor cell line panel (correlation coefficient = 0.68; Supplementary Fig. S2A–S2C).

Figure 2.

The CD44+PrPc+ subpopulation of CRC cells possesses CSC properties. A, double immunofluorescence staining of CD44 (green) and PrPc (red) in a representative CRC cryosection. The section was counterstained with 4′,6-diamidino-2-phenylindole (DAPI; blue). Scale bar, 100 μm. B, a representative 4-color flow cytometric analysis of a CRC. Cells were gated by CD45 negativity and EpCAM positivity to remove leukocytes and stromal cells (left and middle) before the indicated subpopulations were analyzed and sorted out on the basis of PrPc and CD44 expression (right). C, Venn diagram illustrating the degree of overlap between the CD44 and PrPc populations. The percentage of each subpopulation is shown with the median values (M) and ranges indicated. D, clonogenic efficiency of indicated subpopulations freshly isolated from CRC (day 42). Single colonies are magnified to show the detailed morphology. Colonies with diameter ≥0.5 mm were counted for statistical analysis. **, P < 0.001; ***, P < 0.0001. E, representative H&E images of regenerated tumors from CD44+PrPc+ and CD44+PrPc subpopulations in NOD/SCID mice (day 90). Note that the cancer cells from CD44+PrPc+ subpopulation have infiltrated into the underlying muscle tissue. Scale bar, 100 μm. F, growth curve of xenograft tumor initiated from 1 × 106 CD44 RNAi (black), PrPc RNAi (blue), CD44 and PrPc RNAi (bright red) CRC cells in nude mice (n = 6). Parental (brown) and scramble (pink) shRNA cells were used as controls. *, P < 0.01; ***, P < 0.0001.

Figure 2.

The CD44+PrPc+ subpopulation of CRC cells possesses CSC properties. A, double immunofluorescence staining of CD44 (green) and PrPc (red) in a representative CRC cryosection. The section was counterstained with 4′,6-diamidino-2-phenylindole (DAPI; blue). Scale bar, 100 μm. B, a representative 4-color flow cytometric analysis of a CRC. Cells were gated by CD45 negativity and EpCAM positivity to remove leukocytes and stromal cells (left and middle) before the indicated subpopulations were analyzed and sorted out on the basis of PrPc and CD44 expression (right). C, Venn diagram illustrating the degree of overlap between the CD44 and PrPc populations. The percentage of each subpopulation is shown with the median values (M) and ranges indicated. D, clonogenic efficiency of indicated subpopulations freshly isolated from CRC (day 42). Single colonies are magnified to show the detailed morphology. Colonies with diameter ≥0.5 mm were counted for statistical analysis. **, P < 0.001; ***, P < 0.0001. E, representative H&E images of regenerated tumors from CD44+PrPc+ and CD44+PrPc subpopulations in NOD/SCID mice (day 90). Note that the cancer cells from CD44+PrPc+ subpopulation have infiltrated into the underlying muscle tissue. Scale bar, 100 μm. F, growth curve of xenograft tumor initiated from 1 × 106 CD44 RNAi (black), PrPc RNAi (blue), CD44 and PrPc RNAi (bright red) CRC cells in nude mice (n = 6). Parental (brown) and scramble (pink) shRNA cells were used as controls. *, P < 0.01; ***, P < 0.0001.

Close modal

We next asked whether PrPc is associated with CD44 in contributing to CSC properties. The abilities to form anchorage-independent spheres/spheroids in vitro and to regenerate serially transplantable xenograft tumors in immunodeficient mice are 2 of the hallmarks of CSCs (37). We therefore used fluorescence-activated cell sorting (FACS) to purify 4 subpopulations: CD44+PrPc+, CD44+PrPc, CD44PrPc+, and CD44PrPc from patient tumors (Fig. 2B). The purity of each subpopulation was more than 97%, as assessed by post-sort FACS and fluorescence microscopic analyses (data not shown). As expected, CD44+PrPc+ CRC cells displayed significantly greater ability to form colonies compared with CD44+PrPc or CD44PrPc+ cells (P = 0.02 and P = 0.007, respectively), whereas CD44PrPc cancer cells failed to generate colonies (Fig. 2D).

To evaluate their tumorigenic potential, the CD44+PrPc+, CD44+PrPc, CD44PrPc+, and CD44PrPc subpopulations isolated from 9 patients (i.e., cases 22–30 and Supplementary Table S1) were subcutaneously injected intoNOD/SCID mice. As predicted by their clonogenicity data, xenograft tumors were generated from the CD44+PrPc+, CD44+PrPc, and CD44PrPc+ cells, but not from the CD44PrPc cells in 6 of 9 cases (Supplementary Table S2). In limited-dilution assays, the CD44+PrPc+ cells displayed greater tumor-initiating capacity than the other subpopulations (P < 0.001) (Supplementary Table S2). We frequently observed that tumor cells had invaded the underlying muscle tissue in tumors regenerated from CD44+PrPc+ (but not in those regenerated from CD44+PrPc) cells (Fig. 2E). To further test their self-renewal capabilities in vivo, 100 CD44+PrPc+ cancer cells purified from a first-generation xenograft tumor were transplanted into NOD/SCID mice. Our findings indicated that 8 of 9 transplantations grew secondary xenograft tumors that recapitulated the phenotype of the original tumor, and even higher tumorigenic capacities were detected in the third transplantation (Supplementary Fig. S2D and data not shown).

Similar tumor-initiating capacities were also observed in nude mice. We isolated the 4 subpopulations prepurified from CD45 tumor cells, and injected them subcutaneously into nude mice (cases 1–11, Supplementary Table S1). Tumor development was observed in 8 cases (Supplementary Fig. S2E). Again, the CD44+PrPc+ CRC cells exhibited the greatest tumor-initiating capacity (P < 0.05; Supplementary Table S2). As shown in Supplementary Figure S2E, the reconstituted tumors were histologically heterogeneous. PrPc and CD44 coexpression was detected in only a minor subset of cells in regenerated tumors (Supplementary Fig. S2F–S2G), also suggesting that the CD44+PrPc+ cells were capable of self-renewal.

As in leukemic stem cells (15), knocking down CD44 in CRC cells significantly inhibited tumor regeneration (14). To investigate the role of PrPc in tumor initiation, we isolated primary CRC cells from patients and cultured them in ultra-low attachment dishes without serum. Remarkably, PrPc knockdown in the CRC cells also retarded tumor growth in nude mice, although the inhibitory effect was slightly less apparent than with CD44 knockdown (Fig. 2F). However, combined knockdown of both CD44 and PrPc completely abolished tumor development (Fig. 2F). Taken together, these data suggest that the CD44+PrPc+ subpopulation of CRC cells does possess CSC properties.

PrPc+ CSC is more metastatic

Because PrPc is highly expressed in metastatic tissues, we attempted to find out whether it contributes to metastasis in CRC. To recapitulate human cancers with greater fidelity, we transplanted 500, 2,000 and 10,000 cells from each of the CD44+PrPc+, CD44+PrPc, CD44PrPc+ and CD44PrPc subpopulations isolated from 7 patients (i.e., case 24–30, Supplementary Table S1), into the cecal wall of NOD/SCID mice (38). As shown in Table 1, 6 of the 7 patient-derived CRC cells produced cecal tumors, which was consistent with the results obtained from subcutaneously-implanted tumor cells (Supplementary Table S2). The CD44+PrPc+ cells clearly formed tumors with the greatest efficiency (P < 0.01; Table 1 and Supplementary Table S2). Remarkably, macroscopic and microscopic analysis revealed that liver metastases were almost exclusively observed in the mice implanted with PrPc+ cells (i.e., CD44+PrPc+ and CD44PrPc+; Fig. 3A–C; Table 1). Furthermore, semiquantitative measurement of the metastatic area and the metastatic index confirmed that the tumors derived from PrPc+ CRC cells were highly metastatic (Fig. 3D–E).

Figure 3.

PrPc confers greater metastatic capacity. A, 10,000 freshly sorted CD44+PrPc+ cells from CRCs were transplanted into the cecal wall of NOD/SCID mice. The regenerated cecal tumor (black arrow) and the liver metastatic nodules (white arrow) are shown. B, metastatic nodules in the liver formed after injecting 2,000 of the cells into the cecal wall (day 100). Metastatic cancer cells are distinguished from normal mouse liver tissue by having a higher nucleus/cytoplasm ratio. Scale bar, 300 μm. C, metastatic efficiency. The number of cecal tumors with liver metastasis (white bar) vs. cecal tumors without metastasis (black bar) determined 100 days after CRC cell injections. D, metastatic areas were determined by Image-Pro Plus software based on H&E staining. The metastatic area was normalized to CD44+PrPc+ cells. **, P < 0.001. E, PrPc+ cells exhibit a higher metastatic index than PrPc cells. Metastatic lesions were determined under a microscope on the basis of H&E staining, and the metastatic index was defined as the number of metastatic nodules per primary tumor volume. ***, P < 0.0001. F, cell migration assays with the indicated subpopulations freshly isolated from CRCs. Each sample was analyzed in triplicate, and the experiment was carried out using 7 patient tumors. ***, P < 0.0001. G, PrPc and CD44 expression in migrated vs. nonmigrated CRC cells determined by flow cytometry analysis. This experiment was conducted in triplicate using 6 patient tumors.

Figure 3.

PrPc confers greater metastatic capacity. A, 10,000 freshly sorted CD44+PrPc+ cells from CRCs were transplanted into the cecal wall of NOD/SCID mice. The regenerated cecal tumor (black arrow) and the liver metastatic nodules (white arrow) are shown. B, metastatic nodules in the liver formed after injecting 2,000 of the cells into the cecal wall (day 100). Metastatic cancer cells are distinguished from normal mouse liver tissue by having a higher nucleus/cytoplasm ratio. Scale bar, 300 μm. C, metastatic efficiency. The number of cecal tumors with liver metastasis (white bar) vs. cecal tumors without metastasis (black bar) determined 100 days after CRC cell injections. D, metastatic areas were determined by Image-Pro Plus software based on H&E staining. The metastatic area was normalized to CD44+PrPc+ cells. **, P < 0.001. E, PrPc+ cells exhibit a higher metastatic index than PrPc cells. Metastatic lesions were determined under a microscope on the basis of H&E staining, and the metastatic index was defined as the number of metastatic nodules per primary tumor volume. ***, P < 0.0001. F, cell migration assays with the indicated subpopulations freshly isolated from CRCs. Each sample was analyzed in triplicate, and the experiment was carried out using 7 patient tumors. ***, P < 0.0001. G, PrPc and CD44 expression in migrated vs. nonmigrated CRC cells determined by flow cytometry analysis. This experiment was conducted in triplicate using 6 patient tumors.

Close modal
Table 1.

Tumor-initiating incidence and liver metastasis in the orthotopic model

Cecal tumorCD44+PrPc+CD44+PrPcCD44PrPc+CD44PrPc
Case5002,00010,0005002,00010,0005002,00010,0005002,00010,000
24 1/3 1/3 3/3 0/3 1/3 2/3 0/3 1/3 2/3 0/3 0/3 0/3 
25 1/3 2/3 3/3 0/3 1/3 2/3 0/3 1/3 2/3 0/3 0/3 0/3 
26 1/3 2/3 3/3 1/3 1/3 2/3 0/3 0/3 2/3 0/3 0/3 0/3 
27 1/3 1/3 2/3 0/3 1/3 1/3 0/3 0/3 2/3 0/3 0/3 0/3 
28 0/3 1/3 2/3 0/3 0/3 2/3 0/3 0/3 1/3 0/3 0/3 1/3 
29 1/3 1/3 3/3 0/3 1/3 3/3 0/3 1/3 3/3 0/3 0/3 1/3 
Total 5/18 8/18 16/18 1/18 5/18 12/18 0/18 3/18 12/18 0/18 0/18 2/18 
Metastasis CD44+PrPc+ CD44+PrPc CD44PrPc+ CD44PrPc 
Case 500 2,000 10,000 500 2,000 10,000 500 2,000 10,000 500 2,000 10,000 
24 0/3 0/3 2/3 0/3 0/3 0/3 03 1/3 2/3 0/3 0/3 0/3 
25 1/3 1/3 3/3 0/3 0/3 0/3 0/3 1/3 2/3 0/3 0/3 0/3 
26 0/3 0/3 3/3 0/3 0/3 0/3 0/3 0/3 2/3 0/3 0/3 0/3 
27 0/3 0/3 1/3 0/3 0/3 0/3 0/3 0/3 1/3 0/3 0/3 0/3 
28 0/3 0/3 1/3 0/3 0/3 0/3 0/3 0/3 1/3 0/3 0/3 0/3 
29 0/3 1/3 2/3 0/3 1/3 2/3 0/3 1/3 2/3 0/3 0/3 1/3 
Total 1/18 2/18 12/18 0/18 1/18 2/18 0/18 3/18 10/18 0/18 0/18 1/18 
Cecal tumorCD44+PrPc+CD44+PrPcCD44PrPc+CD44PrPc
Case5002,00010,0005002,00010,0005002,00010,0005002,00010,000
24 1/3 1/3 3/3 0/3 1/3 2/3 0/3 1/3 2/3 0/3 0/3 0/3 
25 1/3 2/3 3/3 0/3 1/3 2/3 0/3 1/3 2/3 0/3 0/3 0/3 
26 1/3 2/3 3/3 1/3 1/3 2/3 0/3 0/3 2/3 0/3 0/3 0/3 
27 1/3 1/3 2/3 0/3 1/3 1/3 0/3 0/3 2/3 0/3 0/3 0/3 
28 0/3 1/3 2/3 0/3 0/3 2/3 0/3 0/3 1/3 0/3 0/3 1/3 
29 1/3 1/3 3/3 0/3 1/3 3/3 0/3 1/3 3/3 0/3 0/3 1/3 
Total 5/18 8/18 16/18 1/18 5/18 12/18 0/18 3/18 12/18 0/18 0/18 2/18 
Metastasis CD44+PrPc+ CD44+PrPc CD44PrPc+ CD44PrPc 
Case 500 2,000 10,000 500 2,000 10,000 500 2,000 10,000 500 2,000 10,000 
24 0/3 0/3 2/3 0/3 0/3 0/3 03 1/3 2/3 0/3 0/3 0/3 
25 1/3 1/3 3/3 0/3 0/3 0/3 0/3 1/3 2/3 0/3 0/3 0/3 
26 0/3 0/3 3/3 0/3 0/3 0/3 0/3 0/3 2/3 0/3 0/3 0/3 
27 0/3 0/3 1/3 0/3 0/3 0/3 0/3 0/3 1/3 0/3 0/3 0/3 
28 0/3 0/3 1/3 0/3 0/3 0/3 0/3 0/3 1/3 0/3 0/3 0/3 
29 0/3 1/3 2/3 0/3 1/3 2/3 0/3 1/3 2/3 0/3 0/3 1/3 
Total 1/18 2/18 12/18 0/18 1/18 2/18 0/18 3/18 10/18 0/18 0/18 1/18 

NOTE: Indicated numbers (500, 2,000 and 10,000) of CD44+PrPc+, CD44+PrPc, CD44PrPc+, and CD44PrPc subpopulations freshly isolated from CD45EpCAM+ CRC cells in 7 patients (i.e., case 24–30, Supplementary Table S1) were transplanted into cecal wall of NOD/SCID mice. In 6 of 7 cases, xenograft tumors were regenerated (day 100). The incidence of tumor regeneration at the cecum and the incidence of metastasis to the liver are shown.

The spleen transplantation model is another commonly used approach to test the proliferative and metastatic potential of CSCs (39, 40). CD44+PrPc+, CD44+PrPc, CD44PrPc+, and CD44PrPc subpopulations were sorted from the tumors of 10 patients (i.e., cases 12–21, Supplementary Table S1), and 500 cells from each subpopulation were then transplanted into the spleen of nude mice. Similar to the results obtained using the orthotopic model, liver metastasis occurred within 70 days in 28 of 30 cases of CD44+PrPc+, 3 of 30 cases of CD44+PrPc, 15 of 30 cases of CD44PrPc+, and 2 of 30 cases of CD44PrPc cells (Supplementary Fig. S3A–S3B). These results suggest that PrPc+ is intimately associated with the CRC liver metastasis. Transwell analysis showed that CD44+PrPc+ cells displayed about 16-fold greater migratory ability than CD44+PrPc cells (Fig. 3F). Most of the migrating cells (84.25 ± 3.14%) were found to be PrPc-positive, versus only 1.25 ± 0.22% of the nonmigrant cells (Fig. 3G). Collectively, these observations imply that PrPc is crucial for CCSC migration in vitro and for metastasis in vivo.

PrPc is functionally important for cell migration and cancer metastasis

Because CSC is generally rare and it is difficult to label and trace its growth in animals, we generated a cell line of CD44+ CCSCs isolated from a CRC patient. These cells have all the characteristics of CSCs including constitutive expression of CD44, and the capacity to form spheres and generate xenograft tumors (14). The CCSCs were grown in ultra-low attachment dishes without serum, and were subcultured once a week for fewer than 20 passages. To investigate the unique role of PrPc in metastasis, we used this cell line prelabeled with DsRed and in which PrPc had either been knocked down by shRNAs or was overexpressed via the pBABE-PURO retrovirus system. As shown in Fig. 4A–C, PrPc knockdown strongly inhibited tumor growth and liver metastasis as assessed by whole-body fluorescence imaging and hematoxylin and eosin (H&E) staining, suggesting that PrPc play a central role in CSC metastasis. In contrast, both tumor growth and liver metastasis were promoted by PrPc overexpression. Moreover, the level of PrPc expression was correlated with the metastatic index (Fig. 4D). Similar results were observed using the spleen metastasis model (Supplementary Fig. S4A–S4C). In addition, Transwell migration assays revealed that the levels of PrPc correlated with the migratory capacity of CCSCs (Fig. 4E). It is noteworthy that the PrPc levels did not alter the cell-cycle profiles (Supplementary Fig. S4D). Using CRC cells with CSC properties, we were able to show that PrPc is functionally required for the migration of CCSCs and contributes to CRC metastasis.

Figure 4.

PrPc confers greater metastatic capacity. A, representative whole-body fluorescence imaging of orthotopically growing (dotted circles) and metastasizing CCSCs (n = 9, day 80). B, quantitative imaging data of tumor metastases (day 60). Relative intensity of fluorescence (excluding the cecum region) was calculated. **, P < 0.001; ***, P < 0.0001. C, liver metastasis with manipulated levels of PrPc. Macroimaging of metastatic nodules in liver (top) with boxed regions observed under a stereofluorescence microscope (middle; scale bar, 1 mm). Metastases were verified by H&E staining and indicated by dotted circles (bottom). Scale bar, 100 μm. D, metastatic index of CCSCs with PrPc overexpression or knockdown. Metastatic index was defined as number of metastatic nodules in the liver vs. primary tumor volume. **, P < 0.001; ***, P < 0.0001. E, migratory abilities of CCSCs with PrPc overexpression or knockdown. **, P < 0.001. F, monoclonal anti-PrPc antibodies inhibit the migration of CCSCs in Transwell migration assays. Mouse IgG was used as control. **, P < 0.001; ***, P < 0.0001. G, monoclonal anti-PrPc antibodies inhibit tumor initiation and liver metastasis in vivo (n = 15). Scale bars, 5 mm.

Figure 4.

PrPc confers greater metastatic capacity. A, representative whole-body fluorescence imaging of orthotopically growing (dotted circles) and metastasizing CCSCs (n = 9, day 80). B, quantitative imaging data of tumor metastases (day 60). Relative intensity of fluorescence (excluding the cecum region) was calculated. **, P < 0.001; ***, P < 0.0001. C, liver metastasis with manipulated levels of PrPc. Macroimaging of metastatic nodules in liver (top) with boxed regions observed under a stereofluorescence microscope (middle; scale bar, 1 mm). Metastases were verified by H&E staining and indicated by dotted circles (bottom). Scale bar, 100 μm. D, metastatic index of CCSCs with PrPc overexpression or knockdown. Metastatic index was defined as number of metastatic nodules in the liver vs. primary tumor volume. **, P < 0.001; ***, P < 0.0001. E, migratory abilities of CCSCs with PrPc overexpression or knockdown. **, P < 0.001. F, monoclonal anti-PrPc antibodies inhibit the migration of CCSCs in Transwell migration assays. Mouse IgG was used as control. **, P < 0.001; ***, P < 0.0001. G, monoclonal anti-PrPc antibodies inhibit tumor initiation and liver metastasis in vivo (n = 15). Scale bars, 5 mm.

Close modal

PrPc antibodies inhibit migration and metastasis

Surface markers unique to CSCs would provide ideal therapeutic targets on these cells, and indeed monoclonal antibodies have shown therapeutic efficacy in both preclinical tumor models and patients with cancer (41). We were therefore interested in addressing the question of whether PrPc could be used as a therapeutic target, in a manner similar to CD44 (15). We generated several monoclonal antibodies against PrPc, out of which we selected the IgG1 antibody. The specificity of the antibodies was verified by ELISA, Western blotting (WB), and IHC staining (Supplementary Fig. S4E–S4G). The migration of CD44+ CCSCs in vitro was strongly inhibited after they had been treated with 1 μg/mL anti-PrPc antibodies. The inhibitory effect obtained was greater than that associated with the anti-CD44 antibody (∼1.7 fold). A combination of 1 μg/mL anti-CD44 and 1 μg/mL anti-PrPc antibodies was able to inhibit migration by up to approximately 80% (Fig. 4F); whereas PrPc antibodies did not show any effect on cell migration in PrPc knockdown cells (data not shown).

It was essential to find out whether PrPc monoclonal antibodies can prevent CRC metastases in vivo. NOD/SCID mice bearing orthotopic implantations of CCSCs were administered 200 μg antibodies intraperitoneally on a weekly basis. ELISA analysis revealed that the concentration of anti-PrPc antibody in the serum of tumor-bearing animals was maintained at 0.87 to 1 μg/mL (data not shown). The orthotopic and metastatic tumors were monitored by whole-body fluorescence imaging every week and after 8 to 10 weeks the mice were sacrificed and underwent pathology examinations. The presence of anti-PrPc antibodies dramatically decreased tumorigenicity and liver metastases (Fig. 4G), whereas their bodyweight did not decrease (not shown). CCSCs precoated with anti-PrPc antibodies before orthotopic implantation produced a similar inhibitory effect (Supplementary Fig. S4H).

To make sure that the observed metastasis-inhibiting effects of anti-PrPc antibodies were not due to inhibition of primary tumor growth, we preinoculated the CCSCs orthotopically into the cecal wall for 3 weeks. Mice with equivalent tumors were then grouped and injected with 100 or 200 μg antibodies weekly. The results clearly showed that the anti-PrPc antibody inhibited tumor growth and metastasis at the lower dosage (100 μg/week) and eliminated tumor growth at higher dosage (200 μg/week; Supplementary Fig. S4I).

PrPc promotes EMT in CCSCs

We consistently observed that the PrPc-overexpressing CCSCs exhibited an elongated mesenchymal-like morphology, whereas the PrPc knockdown cells were flatter (Fig. 5A; Supplementary Fig. S4G). Also, PrPc+ CRC cells generated branching colonies with loose cell–cell interaction (see Fig. 2D), characteristic of the EMT phenotype. This prompted us to investigate whether PrPc promotes EMT, a known mechanism of CSC migration and invasion. Cells undergoing EMT retain stem cell functionality and form invasive front colonies (42). We first used quantitative reverse transcription-PCR (qRT-PCR; primers shown in Supplementary Table S3) to measure several EMT-related genes by manipulating PrPc levels. In PrPc overexpressing cells, mesenchymal genes such as N-cadherin and Twist were significantly upregulated, whereas E-cadherin, an epithelial marker, was downregulated (Supplementary Fig. S5A). In addition, there were increased levels of matrix metalloproteinase 2 (MMP2) and MMP11 (Supplementary Fig. S5A), which may have contributed to the metastasis observed in PrPc overexpressing cells. In contrast, knockdown of PrPc resulted in a reversed expression pattern (Supplementary Fig. S5A). Western blot analysis and double immunofluorescence staining further confirmed the correlation of PrPc with the expression of EMT-related molecules (Fig. 5B and C; Supplementary Fig. S5B).

Figure 5.

PrPc modulates EMT in CCSCs. A, phase-contrast images of CCSCs in which PrPc was either overexpressed or knocked down. Note the morphologic changes after the PrPc expression level was modulated. Scale bar, 50 μm. B, Western blot analysis of N-cadherin and E-cadherin in CCSCs in which PrPc was either overexpressed or knocked down. Monoglycosylated (30 kDa) and unglycosylated (28 kDa) PrPc expressed on CCSCs were detected by anti-PrPc (Clone5) antibody. C, double immunofluorescence staining of N-cadherin (green) and E-cadherin (red) in CCSCs in which PrPc was either overexpressed or knocked down. Scale bar, 50 μm. D, qRT-PCR analysis of EMT related genes in Twist or PrPc manipulated CCSCs. The mRNA expression levels were normalized to that of parental CCSCs. Purple bars, knock-in of twist in PrPc RNAi CCSCs. Yellow bars, knock-in of PrPc in Twist RNAi cells. **, P < 0.001; ***, P < 0.0001. E, relative mRNA expression levels of N-cadherin, E-cadherin, and Twist in the indicated subpopulations freshly sorted from patients. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a common control. The mRNA expression levels of each gene were normalized to those of the cell population with the lowest mRNA expression. This experiment was conducted using cells from 6 patient tumors. **, P < 0.001; ***, P < 0.0001. F, representative flow cytometry analysis showing that the N-cadherin-positive CRC cells were also PrPc positive, whereas most E-cadherin-positive cells were PrPc negative. This experiment was carried out in 8 freshly sorted CRCs by gating for CD44EpCAM+.

Figure 5.

PrPc modulates EMT in CCSCs. A, phase-contrast images of CCSCs in which PrPc was either overexpressed or knocked down. Note the morphologic changes after the PrPc expression level was modulated. Scale bar, 50 μm. B, Western blot analysis of N-cadherin and E-cadherin in CCSCs in which PrPc was either overexpressed or knocked down. Monoglycosylated (30 kDa) and unglycosylated (28 kDa) PrPc expressed on CCSCs were detected by anti-PrPc (Clone5) antibody. C, double immunofluorescence staining of N-cadherin (green) and E-cadherin (red) in CCSCs in which PrPc was either overexpressed or knocked down. Scale bar, 50 μm. D, qRT-PCR analysis of EMT related genes in Twist or PrPc manipulated CCSCs. The mRNA expression levels were normalized to that of parental CCSCs. Purple bars, knock-in of twist in PrPc RNAi CCSCs. Yellow bars, knock-in of PrPc in Twist RNAi cells. **, P < 0.001; ***, P < 0.0001. E, relative mRNA expression levels of N-cadherin, E-cadherin, and Twist in the indicated subpopulations freshly sorted from patients. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a common control. The mRNA expression levels of each gene were normalized to those of the cell population with the lowest mRNA expression. This experiment was conducted using cells from 6 patient tumors. **, P < 0.001; ***, P < 0.0001. F, representative flow cytometry analysis showing that the N-cadherin-positive CRC cells were also PrPc positive, whereas most E-cadherin-positive cells were PrPc negative. This experiment was carried out in 8 freshly sorted CRCs by gating for CD44EpCAM+.

Close modal

Twist is a key transcription factor regulating mesenchymal cell fate, differentiation, and morphogenesis (43). Because PrPc expression had been found to influence the expression of Twist (Supplementary Fig. S5A), we reasoned that PrPc might regulate EMT via Twist. To test this, Twist was ectopically expressed in CCSCs in which PrPc was stably knocked down by shRNA. We observed that constitutively overexpressed Twist was indeed able to induce EMT in PrPc-knocked down CCSCs, leading to increased levels of Snail and N-cadherin (Fig. 5D). However, overexpression of PrPc in the Twist RNAi cells was unable to restore EMT, probably because Snail and N-cadherin expression was lower than in the parental cells (Fig. 5D). Overall, we conclude that PrPc regulates the EMT phenotype by modulating Twist.

Next, we used qRT-PCR to measure the mRNA levels of Twist, N-cadherin, and E-cadherin in CD44+PrPc+, CD44+PrPc, CD44PrPc+, and CD44PrPc subpopulations of CRC cells freshly purified from 6 patients. The results showed that both Twist and N-cadherin were expressed at significantly higher levels in the PrPc-positive (i.e. CD44+PrPc+ and CD44PrPc+) CRC cells than in the PrPc subpopulations (Fig. 5E). Furthermore, approximately 82% of N-cadherin+ cells were also PrPc+, whereas more than 93% of E-cadherin+ cells were PrPc in CD45EpCAM+ CRC cells as analyzed by flow cytometry (Fig. 5F). These findings suggest that PrPc is positively correlated with the mesenchymal properties of cells from CRC patients.

Evidence that PrPc-induced EMT is ERK2 dependent

EMT is regulated by the TGF-β and ERK signaling pathways (44). To understand the molecular mechanisms of PrPc-induced EMT, we first assessed the activation of these 2 pathways (45). Knocking down PrPc seemed to result in a reduction of both Phospho-smad2 and ERK1/2; however, overexpression of PrPc activated ERK1/2 but not smad2, suggesting that ERK signaling may act downstream of PrPc (Fig. 6A). In support of this, we found that U0126, an inhibitor of ERK signaling, suppressed the expression of several EMT-associated molecules (Fig. 6B) as well as EMT itself, especially when PrPc was overexpressed.

Figure 6.

PrPc-regulated EMT is ERK2 dependent. A, Western blot analysis of the proteins indicated in CCSCs in which PrPc was either overexpressed or knocked down. B, parental CCSCs and those overexpressing PrPc were treated with the ERK2 inhibitor U0126 for 24 hours, and a whole-cell lysate was used for Western Blot analysis of the molecules indicated. C, CCSCs were treated with 5 μg/mL anti-PrPc antibody (Ab), 2 μg/ml laminin (LN), and a combination of Ab and LN for 24 hours, after which cells were harvested for Western Blot analysis. D, phase-contrast images of CCSCs maintained under the conditions indicated. Note that the anti-PrPc antibodies rescued the morphologic changes in the presence of LN. Scale bar, 100 μm. E, Double immunofluorescence staining of N-cadherin (green) and E-cadherin (red) in CCSCs in the presence of 10 μmol/L U0126, 2 μg/mL LN, and indicated dose of anti-PrPc antibodies (μg/mL). Scale bar, 20 μm. F, migration of CCSCs in the presence of U0126, 2 μg/mL LN, and indicated dose of anti-PrPc antibodies (μg/mL).

Figure 6.

PrPc-regulated EMT is ERK2 dependent. A, Western blot analysis of the proteins indicated in CCSCs in which PrPc was either overexpressed or knocked down. B, parental CCSCs and those overexpressing PrPc were treated with the ERK2 inhibitor U0126 for 24 hours, and a whole-cell lysate was used for Western Blot analysis of the molecules indicated. C, CCSCs were treated with 5 μg/mL anti-PrPc antibody (Ab), 2 μg/ml laminin (LN), and a combination of Ab and LN for 24 hours, after which cells were harvested for Western Blot analysis. D, phase-contrast images of CCSCs maintained under the conditions indicated. Note that the anti-PrPc antibodies rescued the morphologic changes in the presence of LN. Scale bar, 100 μm. E, Double immunofluorescence staining of N-cadherin (green) and E-cadherin (red) in CCSCs in the presence of 10 μmol/L U0126, 2 μg/mL LN, and indicated dose of anti-PrPc antibodies (μg/mL). Scale bar, 20 μm. F, migration of CCSCs in the presence of U0126, 2 μg/mL LN, and indicated dose of anti-PrPc antibodies (μg/mL).

Close modal

Laminin is a PrPc ligand (46), which, we reasoned, could activate events occurring downstream of PrPc, such as EMT. Laminin did enhance both ERK2 phosphorylation and the EMT phenotype (Fig. 6C and D). Interestingly, PrPc antibody blocked the laminin-induced ERK activation, EMT protein expression and EMT morphology in a dose-dependent manner, (Fig. 6C and D), probably via interference with the interaction between laminin and its receptor PrPc. Immunofluorescence staining confirmed that the anti-PrPc antibody attenuated the laminin-induced N-cadherin expression (Fig. 6E). In addition, Transwell migration assays revealed that U0126 blocked the PrPc overexpression-induced migration, and that the anti-PrPc antibody also attenuated laminin-enhanced migration (Fig. 6F). Our findings thus suggest that ERK2 signaling pathway is involved in PrPc-induced EMT.

It has been shown that CSCs contribute to both tumor initiation and cancer metastasis in many tumor systems. However, in most cases, the underlying mechanisms are not well understood. In this study, we have identified a distinct PrPc+ subpopulation in CD44+ CCSCs that displays enhanced capacity to regenerate tumors. These results emphasize the functionalities of PrPc as an “enrichment” factor for CCSC. Our data show that PrPc is a novel CCSC biomarker, which, by itself, functionally contributes to both CRC tumorigenicity and metastasis. Interestingly, the PrPc+ CRC cells tend to localize at the invasion fronts in patient tumors, are more mesenchymal, and behave like other cancer cells undergoing EMT, properties that have probably given these cells a greater capacity to disseminate and metastasize. These observations indicate PrPc's role in regulating CCSC likely occurring through its ability to promote EMT and the exact mechanism warrants future investigations. In terms of the mechanism involved, our work implies that PrPc may regulate EMT through ERK2- and Twist-dependent signaling pathways. Importantly, the interaction of PrPc with its ligand, laminin, directly activates ERK2 signaling and promotes EMT. Therefore, conceptually, interference with PrPc expression/functions, the PrPc-laminin interaction or its downstream signaling events, may be a promising avenue in developing new therapies targeting PrPc+ CRC cell-mediated metastasis.

Previous work has shown that PrP1, a PrPc ortholog in the zebra fish, regulates the stability of E-cadherin mediated cell adhesion (47), suggesting that PrPc expression may endow the cell with properties of EMT, which is a developmentally conserved regulatory mechanism. In addition, PrPc-null (PRNP0/0) cells display less migratory activity than PRNP+/+ cells (48). The more aggressive phenotype of PRNP+/+ cells is consistent with our observations that PrPc contributes to CCSC migration. Recently, CD26+ CRC cells have also been found to possess CSC properties, with greatly increased metastatic capacity, and, intriguingly, like PrPc, CD26 also functions as an adhesion molecule that binds to ECM (24). These findings, together with our present study, strongly suggest that the enhanced tumorigenic and metastatic capacities of CCSCs are probably determined by the CSC-intrinsic expression of functional molecules such as PrPc and CD26, which in turn mediate the interactions of CSCs with their microenvironment. Future work will attempt to elucidate the potential relationship between PrPc+ and CD26+ CCSCs.

Previous studies have shown that the expression of PrPc is positively correlated with invasiveness and drug resistance in both gastric and breast cancers (27, 49). In addition, Liao and colleagues have reported that endoglin and PrPc are enriched in mammosphere-initiating cells (50). Although PrPc has been reported to be highly expressed in some CRCs (51), as far as we are aware this is the first report linking PrPc to CCSCs, and our data help to explain how PrPc is associated with cancer metastasis. The level of PrPc was inversely correlated with patient survival, suggesting a potential prognostic value in CRC. It has recently emerged that monoclonal antibodies targeting surface molecules expressed in CSCs show powerful therapeutic efficacy by eliminating CSCs and leading to tumor resolution (16). Our findings therefore have significant clinical implications, as PrPc+ CSCs may provide important new targets for therapy in metastatic CRC. Indeed, PrPc monoclonal antibodies display strong inhibitory effects on CCSC tumorigenesis and metastasis in orthotopic xenograft models, suggesting possible therapeutic applications in the clinic. Overall, our findings throw new light on the mechanisms underlying CRC cell metastasis and are of clinical importance for the design of therapeutic strategies for treating CRC.

No potential conflicts of interest were disclosed.

Conception and design: L. Du, B. Laffin, M. Mehrpour, Q. Chen

Development of methodology: L. Du, G. Rao, B. Li, Y. Zhu, D.G. Tang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Rao, H. Wang, B. Li, W. Tian, J.T. Cui, L. He, X. Tian, C. Hao, H. Liu, Y. Zhu, D.G. Tang, Y. Lu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Du, G. Rao, H. Wang, B. Li, B. Laffin, D.G. Tang, M. Mehrpour

Writing, review, and/or revision of the manuscript: L. Du, G. Rao, B. Li, M. Mehrpour, Y. Lu, Q. Chen

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): G. Rao, B. Li, X. Sun, D.G. Tang

Study supervision: Y. Lu, Q. Chen

The authors would like to thank Drs. G. Ge and K.-J. Wu for providing Twist plasmids, Dr. S. L. Holbeck for providing CMA slides, Dr. J. Grassi for providing antibody against PrPc (SAF69), Mrs. X. Zhang, T. Zhao, J. Wang, and L. Dou for FACS assistance, Mrs. Y. Zhou and X. Shi for whole-body fluorescence imaging assays, Dr. W. Zhang and Dr. C. Han for biostatistical analysis, Dr. A. Lu, Dr. W. Xue, and Dr. L. Zhou for pathologic determination, and Dr. A. Zhou for his suggestions and revision of our manuscript.

This work was supported by initiative of stem cell and regenerative medicine from Chinese Academy of Sciences (XDA01040409), the 973 project from the Ministry of Science and Technology (MOST) of China (2009CB512800 and 2010CB912204), and a grant from the National Natural Science Foundation of China (NSFC; 81130045 to Q. Chen and 31000614 to L. Du), from MOST 2010CB529403 (X. Tian), MOST 2010CB912204 (Y. Zhu), and from Institut National de la Santé et de la Recherche Médicale, Partenariat Hubert Curien - Programme Français de Coopération avec la Chine, Fondation Franco-Chinoise pour la Science et ses Applications (M. Mehrpour).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Sporn
MB
. 
The war on cancer
.
Lancet
1996
;
347
:
1377
81
.
2.
Jemal
A
,
Thun
MJ
,
Ward
EE
,
Henley
SJ
,
Cokkinides
VE
,
Murray
TE
. 
Mortality from leading causes by education and race in the United States, 2001
.
Am J Prev Med
2008
;
34
:
1
8
.
3.
Pardal
R
,
Clarke
MF
,
Morrison
SJ
. 
Applying the principles of stem-cell biology to cancer
.
Nat Rev Cancer
2003
;
3
:
895
902
.
4.
Bonnet
D
,
Dick
JE
. 
Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell
.
Nat Med
1997
;
3
:
730
7
.
5.
Al-Hajj
M
,
Wicha
MS
,
Benito-Hernandez
A
,
Morrison
SJ
,
Clarke
MF
. 
Prospective identification of tumorigenic breast cancer cells
.
Proc Natl Acad Sci U S A
2003
;
100
:
3983
8
.
6.
Singh
SK
,
Clarke
ID
,
Terasaki
M
,
Bonn
VE
,
Hawkins
C
,
Squire
J
, et al
Identification of a cancer stem cell in human brain tumors
.
Cancer Res
2003
;
63
:
5821
8
.
7.
Kim
CF
,
Jackson
EL
,
Woolfenden
AE
,
Lawrence
S
,
Babar
I
,
Vogel
S
, et al
Identification of bronchioalveolar stem cells in normal lung and lung cancer
.
Cell
2005
;
121
:
823
35
.
8.
Dalerba
P
,
Dylla
SJ
,
Park
IK
,
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
.
9.
Ricci-Vitiani
L
,
Lombardi
DG
,
Pilozzi
E
,
Biffoni
M
,
Todaro
M
,
Peschle
C
. 
Identification and expansion of human colon-cancer-initiating cells
.
Nature
2007
;
445
:
111
5
.
10.
O'Brien
CA
,
Pollett
A
,
Gallinger
S
,
Dick
JE
. 
A human colon cancer cell capable of initiating tumour growth in immunodeficient mice
.
Nature
2007
;
445
:
106
10
.
11.
Huang
EH
,
Hynes
MJ
,
Zhang
T
,
Ginestier
C
,
Dontu
G
,
Appelman
H
, et al
Aldehyde dehydrogenase 1 is a marker for normal and malignant human colonic stem cells (SC) and tracks SC overpopulation during colon tumorigenesis
.
Cancer Res
2009
;
69
:
3382
9
.
12.
Takeda
K
,
Kinoshita
I
,
Shimizu
Y
,
Matsuno
Y
,
Shichinohe
T
,
Dosaka-Akita
H
. 
Expression of LGR5, an Intestinal Stem Cell Marker, During Each Stage of Colorectal Tumorigenesis
.
Anticancer Res
2011
;
31
:
263
70
.
13.
Vermeulen
L
,
Todaro
M
,
de Sousa
MF
,
Sprick
MR
,
Kemper
K
,
Perez
AM
, et al
14. 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
.
14.
Du
L
,
Wang
H
,
He
L
,
Zhang
J
,
Ni
B
,
Wang
X
, et al
CD44 is of functional importance for colorectal cancer stem cells
.
Clin Cancer Res
2008
;
14
:
6751
60
.
15.
Jin
L
,
Hope
KJ
,
Zhai
Q
,
Smadja-Joffe
F
,
Dick
JE
. 
Targeting of CD44 eradicates human acute myeloid leukemic stem cells
.
Nat Med
2006
;
12
:
1167
74
.
16.
Majeti
R
. 
Monoclonal antibody therapy directed against human acute myeloid leukemia stem cells
.
Oncogene
2011
;
30
:
1009
19
.
17.
Dalerba
P
,
Clarke
MF
. 
Cancer stem cells and tumor metastasis: first steps into uncharted territory
.
Cell Stem Cell
2007
;
1
:
241
2
.
18.
Wicha
MS
. 
Cancer stem cells and metastasis: lethal seeds
.
Clin Cancer Res
2006
;
12
:
5606
7
.
19.
Hermann
PC
,
Huber
SL
,
Herrler
T
,
Aicher
A
,
Ellwart
JW
,
Guba
M
, et al
Distinct populations of cancer stem cells determine tumor growth and metastatic activity in human pancreatic cancer
.
Cell Stem Cell
2007
;
1
:
313
23
.
20.
Gao
AC
,
Lou
W
,
Dong
JT
,
Isaacs
JT
. 
CD44 is a metastasis suppressor gene for prostatic cancer located on human chromosome 11p13
.
Cancer Res
1997
;
57
:
846
9
.
21.
Bankfalvi
A
,
Krassort
M
,
Buchwalow
IB
,
Vegh
A
,
Felszeghy
E
,
Piffko
J
. 
Gains and losses of adhesion molecules (CD44, E-cadherin, and beta-catenin) during oral carcinogenesis and tumour progression
.
J Pathol
2002
;
198
:
343
51
.
22.
Jothy
S
. 
CD44 and its partners in metastasis
.
Clin Exp Metastasis
2003
;
20
:
195
201
.
23.
Martin
TA
,
Harrison
G
,
Mansel
RE
,
Jiang
WG
. 
The role of the CD44/ezrin complex in cancer metastasis
.
Crit Rev Oncol Hematol
2003
;
46
:
165
86
.
24.
Pang
R
,
Law
WL
,
Chu
AC
,
Poon
JT
,
Lam
CS
,
Chow
AK
, et al
A subpopulation of CD26+ cancer stem cells with metastatic capacity in human colorectal cancer
.
Cell Stem Cell
2010
;
6
:
603
15
.
25.
Bueler
H
,
Aguzzi
A
,
Sailer
A
,
Greiner
RA
,
Autenried
P
,
Aguet
M
, et al
Mice devoid of PrP are resistant to scrapie
.
Cell
1993
;
73
:
1339
47
.
26.
Klein
MA
,
Aguzzi
A
. 
The neuroimmune interface in prion diseases
.
News Physiol Sci
2000
;
15
:
250
5
.
27.
Mehrpour
M
,
Codogno
P
. 
Prion protein: from physiology to cancer biology
.
Cancer Lett
2010
;
290
:
1
23
.
28.
Aguzzi
A
,
Polymenidou
M
. 
Mammalian prion biology: one century of evolving concepts
.
Cell
2004
;
116
:
313
27
.
29.
Dodelet
VC
,
Cashman
NR
. 
Prion protein expression in human leukocyte differentiation
.
Blood
1998
;
91
:
1556
61
.
30.
Zhang
CC
,
Steele
AD
,
Lindquist
S
,
Lodish
HF
. 
Prion protein is expressed on long-term repopulating hematopoietic stem cells and is important for their self-renewal
.
Proc Natl Acad Sci U S A
2006
;
103
:
2184
9
.
31.
Tang
Z
,
Zhao
M
,
Ji
J
,
Yang
G
,
Hu
F
,
He
J
, et al
Overexpression of gastrin and c-met protein involved in human gastric carcinomas and intestinal metaplasia
.
Oncol Rep
2004
;
11
:
333
9
.
32.
Wang
HY
,
Zhang
JY
,
Cui
JT
,
Tan
XH
,
Li
WM
,
Gu
J
, et al
Expression status of S100A14 and S100A4 correlates with metastatic potential and clinical outcome in colorectal cancer after surgery
.
Oncol Rep
2010
;
23
:
45
52
.
33.
Ki
DH
,
Jeung
HC
,
Park
CH
,
Kang
SH
,
Lee
GY
,
Lee
WS
, et al
Whole genome analysis for liver metastasis gene signatures in colorectal cancer
.
Int J Cancer
2007
;
121
:
2005
12
34.
Koinuma
K
,
Yamashita
Y
,
Liu
W
,
Hatanaka
H
,
Kurashina
K
,
Wada
T
, et al
Epigenetic silencing of AXIN2 in colorectal carcinoma with microsatellite instability
.
Oncogene
2006
;
25
:
139
46
.
35.
Smith
JJ
,
Deane
NG
,
Wu
F
,
Merchant
NB
,
Zhang
B
,
Jiang
A
, et al
Experimentally derived metastasis gene expression profile predicts recurrence and death in patients with colon cancer
.
Gastroenterology
2010
;
138
:
958
68
36.
Wellner
U
,
Schubert
J
,
Burk
UC
,
Schmalhofer
O
,
Zhu
F
,
Sonntag
A
, et al
The EMT-activator ZEB1 promotes tumorigenicity by repressing stemness-inhibiting microRNAs
.
Nat Cell Biol
2009
;
11
:
1487
95
.
37.
Malanchi
I
,
Santamaria-Martínez
A
,
Susanto
E
,
Peng
H
,
Lehr
HA
,
Delaloye
JF
, et al
Interactions between cancer stem cells and their niche govern metastatic colonization
.
Nature
2012
;
481
:
85
9
.
38.
Fu
X
,
Besterman
JM
,
Monosove
A
,
Hoffman
RM
. 
Models of human metastatic colon cancer in nude mice orthotopically constructed by using histologically intact patient specimens
.
Proc Natl Acad Sci U S A
1991
;
88
:
9345
9
.
39.
Brand
MI
,
Casillas
S
,
Dietz
DW
,
Milsom
JW
,
Vladisavljevic
A
. 
Development of a reliable colorectal cancer liver metastasis model
.
J Surg Res
1996
;
63
:
425
32
.
40.
Giavazzi
R
,
Jessup
JM
,
Campbell
DE
,
Walker
SM
,
Fidler
IJ
. 
Experimental nude mouse model of human colorectal cancer liver metastases
.
J Natl Cancer Inst
1986
;
77
:
1303
8
.
41.
Chao
MP
,
Alizadeh
AA
,
Tang
C
,
Myklebust
JH
,
Varghese
B
,
Gill
S
, et al
Anti-CD47 antibody synergizes with rituximab to promote phagocytosis and eradicate non-Hodgkin lymphoma
.
Cell
2010
;
142
:
699
713
.
42.
Brabletz
T
,
Jung
A
,
Spaderna
S
,
Hlubek
F
,
Kirchner
T
. 
Opinion: migrating cancer stem cells - an integrated concept of malignant tumour progression
.
Nat Rev Cancer
2005
;
5
:
744
9
.
43.
Yang
J
,
Mani
SA
,
Donaher
JL
,
Ramaswamy
S
,
Itzykson
RA
,
Come
C
, et al
Twist, a master regulator of morphogenesis, plays an essential role in tumor metastasis
.
Cell
2004
;
117
:
927
39
.
44.
Zavadil
J
,
Böttinger
EP
. 
TGF-b and epithelial-to-mesenchymal transitions
.
Oncogene
2005
;
24
:
5764
74
.
45.
Shin
S
,
Dimitri
CA
,
Yoon
SO
,
Dowdle
W
,
Blenis
J
. 
ERK2 but not ERK1 induces epithelial-to-mesenchymal transformation via DEF motif-dependent signaling events
.
Mol Cell
2010
;
38
:
114
27
.
46.
Rieger
R
,
Edenhofer
F
,
Lasmezas
CI
,
Weiss
S
. 
The human 37-kDa laminin receptor precursor interacts with the prion protein in eukaryotic cells
.
Nat Med
1997
;
3
:
1383
8
.
47.
Malaga-Trillo
E
,
Sempou
E
. 
PrPs: Proteins with a purpose: Lessons from the zebrafish
.
Prion
2009
;
3
:
129
33
.
48.
Muras
AG
,
Hajj
GN
,
Ribeiro
KB
,
Nomizo
R
,
Nonogaki
S
,
Chammas
R
, et al
Prion protein ablation increases cellular aggregation and embolization contributing to mechanisms of metastasis
.
Int J Cancer
2009
;
125
:
1523
31
.
49.
Pan
Y
,
Zhao
L
,
Liang
J
,
Liu
J
,
Shi
Y
,
Liu
N
, et al
Cellular prion protein promotes invasion and metastasis of gastric cancer
.
FASEB J
2006
;
20
:
1886
8
.
50.
Liao
MJ
,
Zhang
CC
,
Zhou
B
,
Zimonjic
DB
,
Mani
SA
,
Kaba
M
, et al
Enrichment of a population of mammary gland cells that form mammospheres and have in vivo repopulating activity
.
Cancer Res
2007
;
67
:
8131
8
.
51.
Antonacopoulou
AG
,
Palli
M
,
Marousi
S
,
Dimitrakopoulos
FI
,
Kyriakopoulou
U
,
Tsamandas
AC
, et al
Prion protein expression and the M129V polymorphism of the PRNP gene in patients with colorectal cancer
.
Mol Carcinog
2010
;
49
:
693
9
.