Background
Colorectal cancer (CRC) is among the three leading causes of cancer-related deaths worldwide. Nearly 50% of patients with CRC develop liver metastases synchronously or metachronously, and in advanced disease the mortality of CRC is principally attributable to the development of hepatic metastases [
1,
2]. Therefore, it is important to uncover the biological mechanisms underlying liver metastasis of CRC and accelerate the development of new treatment strategies.
Cancer stem cells (CSCs) have moved to the center stage in cancer research in recent years and have been viewed as the origin of cancer formation, development and metastasis. CSCs possess the ability to self-renew and to differentiate into phenotypically diverse progeny, a subpopulation within a tumor that could also be labeled tumor-initiating cells [
3‐
5]. Investigation into hematopoietic stem cells has led the way for CSC research [
6], and has been followed by studies showing the existence of CSCs in various types of tumors, including colon cancer [
7‐
12]. Recently, Brabletz and colleagues proposed a concept that CSCs may represent a heterogeneous population consisting of two forms of CSCs during tumor progression, namely stationary and migrating CSCs. The latter is a small subpopulation that combines the two most decisive traits, stemness and mobility, and thus holds important clues for the further understanding of malignant progression [
13].
Recent studies have highlighted the role of chemokines in cancer metastasis. According to the signaling/homing theory, target organs produce and release specific chemokines and attract nearby or distant cancer cells bearing corresponding receptors [
14]. These studies have suggested that the stromal cell-derived factor-1 (SDF-1/CXCR4) axis plays a key role in tumor invasiveness leading to local progression and tumor metastasis in lung, pancreatic, and breast cancers, as well as CRCs [
15‐
20]. Hermann
et al. found that in human pancreatic cancers, a distinct subpopulation of CD133
+CXCR4
+ CSCs was identified that determines the metastatic phenotype of the individual tumor. Depletion of this specific stem cell population virtually abrogated the tumor metastatic phenotype without affecting their tumorigenic potential [
21]. However, the existence of a migrating subpopulation expressing CD133 and CXCR4 has not been reported in CRC.
The acquisition of the mesenchymal phenotype by epithelial cells, known as the epithelial-mesenchymal transition (EMT), is a key process that is required during embryonic development. Epithelial cells have tight cell-cell contact
via various junctions, which only allow limited movement of epithelial cells. In contrast, with an elongated spindle shape, mesenchymal cells interact with neighboring cells to a limited extent (and only at focal points) and have increased motility [
22,
23]. EMT is associated with cancer cell migration and metastasis, and cancer cells acquire a more aggressive phenotype
via EMT, indicating that it is a crucial event in malignancy [
24‐
27]. Some studies have reported a correlation between CSCs and EMT [
27‐
30]. We hypothesized that EMT plays an essential role in endowing migratory CSCs with metastatic capacity. In this study, we have provided evidence for the existence of a distinct migrating CSC subpopulation of CD133
+CXCR4
+ cells in human CRC specimens as well as in the human colon cancer cell line, HCT116. We found that EMT and the SDF-1/CXCR4 axis are involved in the metastatic process.
Methods
Tissue samples
Primary CRC and metastatic liver cancer tissue samples were obtained from 29 patients undergoing surgical resection of primary CRC and/or liver metastasis at the Department of Surgery, Changhai Hospital and Eastern Hepatobiliary Surgery Hospital of the Second Military Medical University from February 2007 to May 2008. After resection, patients were followed up every three months. Sections were reviewed by two experienced pathologists to verify the histologic assessment. All the specimens were adenocarcinoma. Prior informed consent was obtained and the study protocol was approved by the Ethics Committee of the Second Military Medical University.
Cell culture and animals
The human colon cancer cell line, HCT116, was maintained in McCoy's 5A Medium (GIBCO, Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS; GIBCO, Invitrogen), 100 units/ml penicillin and 100 mg/ml streptomycin in a humidified incubator under 95% air and 5% CO2 at 37°C.
Male nude mice (BALB/c strain), six to eight weeks old, were purchased from the Shanghai Experimental Animal Center of the Chinese Academy of Sciences (Shanghai, China). Mice in this study were housed under pathogen-free conditions, and all procedures were performed in accordance with the institutional animal welfare guidelines of the Second Military Medical University.
Flow cytometry and FACS
Fresh specimens from primary CRC, hepatic metastatic cancer and their corresponding normal tissues were transferred to a petri dish, where the tissue was gently minced and filtered (100 mm) to remove large aggregates. This was followed by incubation for 45 minutes at 37°C in 50 ml of Hank's balanced salt solution containing 0.05% collagenase, with continuous stirring. DNAase (0.5 mg) in 1.0 ml of PBS was added 20 to 40 minutes after this incubation period. The cell suspension was filtered (40 mm), and non-parenchymal cells were separated by discontinuous density gradients of Percoll (Pharmacia Biotech, Piscataway, NJ, USA) at 1.044 g/ml and 1.07 g/ml. The final cell suspension was washed twice, and CD133 (Miltenyi Biotech, Bergisch Gladbach, Germany) and/or CXCR4 antibody (eBioscience, San Diego, CA, USA) was added and incubated at 4°C for 20 minutes before washing. Stained cells were analyzed using flow cytometry.
The CD133+CXCR4+ cancer cell content determined by flow cytometry was utilized to investigate the correlation between CD133+CXCR4+ cancer cells and clinical characteristics and two-year survival. Suspensions of HCT-116 cells (107/ml) were sorted according to the expression of CD133 and CXCR4 with a fluorescence activated cell sorting system (FACS, Becton Dickinson, San Jose, CA, USA) following multicolor staining as described for flow cytometric analyses. Separated subpopulations were reanalyzed for purity and then used in subsequent experiments.
Tumor cells (5 × 10
5) were injected into the lateral tail vein using a 27-gauge needle, more experimental details were performed as previously described [
17]. At 120 days post-injection, mice were sacrificed and tissues were examined macroscopically and microscopically for occurrence of metastases.
Clonogenic assay
About 5 × 102 cells were added into each well of a six-well culture plate (three wells for each group). After incubation at 37°C for 14 days, the cells were washed twice with PBS and stained with 0.1% crystal violet solution. The number of colonies containing ≥20 cells was counted under a microscope.
Subcutaneous tumorigenic assay
Subcutaneous administration of colon tumor cells was performed in the armpit area of nude mice. Approximately 1 × 106 cells were injected at each site. Mice were killed 30 days later, and tumorigenic incidence was assessed. The xenografts were excised for weight evaluation.
Real-time RT-PCR
After FACS isolation, cells (3 × 10
5 cells per well) were cultured in six-well plates to 50% to 60% confluence. The treatment group was subjected to SDF-1 (Peprotech, Rocky Hill, NJ, USA) at a concentration of 100 ng/ml for 12 hours. The cells were collected to extract total cellular mRNA with Trizol reagent (Invitrogen, Carlsbad, CA, USA). Expression of mRNA was determined by real-time RT-PCR using SYBR Green Master Mix (Applied Biosystems, Foster City, CA, USA). Total sample RNA was normalized to endogenous GADPH mRNA. The sequences of primers used in this study are shown in Table
1. Thermal cycling conditions included an initial hold period at 95°C for four minutes; this was followed by a two-step PCR program of 95°C for 20 seconds and 72°C for 30 seconds repeated for 40 cycles on an Mx4000 system (Stratagene, La Jolla, CA, USA).
Table 1
Oligonucleotide sequences of primers used in real-time RT-PCR.
E-cadherin | F | TGAAGGTGACAGAGCCTCTGGA |
| R | TGGGTGAATTCGGGCTTGTT |
Vimentin | F | TGGCCGACGCCATCAACACC |
| R | CACCTCGACGCGGGCTTTGT |
N-cadherin | F | GCGCGTGAAGGTTTGCCAGTG |
| R | CCGGCGTTTCATCCATACCACAA |
β-catenin | F | AGCCGACACCAAGAAGCAGAGATG |
| R | CGGCGCTGGGTATCCTGATGT |
Snail | F | CCTCCCTGTCAGATGAGGAC |
| R | CCAGGCTGAGGTATTCCTTG |
GAPDH | F | TGCCAAATATGATGACATCAAGAA |
| R | GGAGTGGGTGTCGCTGTTG |
Boyden chamber invasion assay
A Boyden chamber was separated into two compartments by a polycarbonate membrane with an 8-mm pore, over which a thin layer of extracellular matrix (ECM) was dried. The ECM layer occluded membrane pores, blocking noninvasive cells from migrating. The Boyden chamber invasion assay was performed as previously described [
31]. For the experiment that did not require SDF-1 treatment, 1 × 10
5 cancer cells in 200 μl of serum-free medium were added to the top chamber. McCoy's 5A Medium containing 10% FBS was added to the lower chamber. For the experiment subjected to SDF-1 treatment, both upper and lower chambers were filled with McCoy's 5A Medium containing 1% FBS for the control group, while SDF-1 at a concentration of 100 ng/ml was added to the lower chamber for the treatment group. After incubation for 48 hours, the noninvasive cells were removed with a cotton swab. The cells that had migrated through the membrane and adhered to the lower surface of the membrane were fixed with methanol for ten minutes and stained with crystal violet solution (0.1%). For quantification, the cells were counted using a microscope from five randomized fields at ×200 magnification.
Transwell cell migration assays
Transwell cell migration assays were performed using a protocol similar to that used for the invasive assay described above. A Boyden chamber lacking a thin layer of ECM and a higher density of cells (2.5 × 105 cells) was used.
Cells (5 × 10
5) were intrasplenically administered, and five minutes later the spleen was resected and more experiments were performed as previously described [
32]. The three groups for injection were: CD133
+CXCR4
- cells; CD133
+CXCR4
+ cells; and CD133
+CXCR4
+ cells with AMD3100 (Sigma, St. Louis, MO, USA) administration. AMD3100 (2.5 mg/kg) or PBS was intraperitoneally administered twice a day over 20 days. Mice were sacrificed 45 days later and livers were harvested to observe metastatic tumor formation.
Statistical analysis
All of the
in vitro experiments were repeated at least three times. The data were expressed as means ± SD. Statistical analysis was performed by Student's t test. The comparison of the incidence of tumor/metastasis formation by
in vivo mouse models was performed with Fisher's exact test. For analyzing the association between CD133
+CXCR4
+ cell content and various clinical factors, the 'age' as a continuous variable was expressed as mean (SD) and comparison between groups was done by using the Student's t-test. Categorical variables including gender, location, N status and M status were analyzed by Fisher's exact test and ranked variables including TNM (tumor-node-metastasis) staging, T status and grading were analyzed with the permutation test . The Kaplan-Meier method was used to estimate the medians for time-to-event parameters and to construct the survival curve. The equality of the two curves was compared by the permutation test. The permutation test was conducted by randomly permuting samples' labels (for example, high versus low CD133
+CXCR4
+ cell content) and recomputing the two-sample statistic (for example, Log Rank χ2 for survival) for 50,000 times. The permutation
P value was determined by the proportion of random permuted data sets which resulted in the same or more extreme statistic as observed in the actual data [
33‐
35]. The permutation test was performed using SAS and all other analyses were performed using SPSS, version 17.0 (SPSS Inc, Chicago, Illinois) and tests were two-sided with a significance level <0.05 [
36,
37].
Discussion
CD133 has been used as a marker of tumor-initiating cells in neural cancers and is also generally accepted as a CSC marker for colon cancer [
10‐
12]. However, there are some reports suggesting that CD133
+ cancer cells are not a true representation of CSCs in colon cancer [
39,
40]. We found that CD133
+ colon cancer cells isolated from the HCT116 cell line had a greater clonogenic and tumorigenic ability than CD133
- cells irrespective of CXCR4 expression. The
in vitro and
in vivo assays lend credence to the viewpoint that CD133 could be a marker for colon cancer tumor-initiating cells.
In 2005, Brabletz
et al. proposed the concept that there are two forms of CSCs in tumor progression, namely stationary CSCs and migratory CSCs [
13]. Hermann and colleagues published data supporting the existence of these two distinct subsets in CD133
+ pancreatic CSCs. The CSCs co-expressing CXCR4 were cancer cells with a migratory and invasive phenotype in pancreatic cancer [
21]. In specimens from CRC patients,Pang
et al. demonstrated the existence of migratory CSCs with the CD26 surface antigen as a marker [
41]. In this study, we determined that the percentage of CD133
+CXCR4
+ cancer cells in metastatic liver tumors was nearly eight times higher than that in primary colorectal tumors, indicating enrichment of this CSC subpopulation in metastatic liver tumors and their potential involvement in CRC metastasis to the liver. Transwell migration and invasion assay results indicated that the CD133+CXCR4+ subpopulation had higher migratory and invasive capacities
in vitro. Consistent results were obtained by the standard tail vein metastatic assay
in vivo. This indicated that CD133
+CXCR4
+ cancer cells are a subpopulation of CSCs with a metastatic phenotype. To evaluate the metastatic capacity of different subpopulations, we employed the tail vein metastasis model, which is also known as the experimental metastasis model. The limitation of this model lies in the fact that it cannot reflect the complete metastatic process as does the spontaneous metastasis model in which the tumor cells are injected into the liver and allowed to first form a primary tumor. The complete metastasis cascade includes the following steps: escape of cells from the primary tumor, entry of cells into the lymphatic or blood circulation (intravasation), survival and transport in circulation, escape of cells from circulation (extravasation), and growth of cells to form secondary tumors in a new organ environment [
42]. However, the tail vein metastasis model is able to mimic the extravasation of cancer cells from blood vessels in the target organ which is regarded as a critical step in the metastatic process[
43]. Therefore, as in many studies[
17,
44,
45], it is sufficient to use this model for the comparison of metastatic capacity among different groups.
EMT results in morphological and molecular changes that occur when epithelial cells lose their characteristics and gain mesenchymal properties. The expression of mesenchymal markers, such as N-cadherin and vimentin, and the loss of E-cadherin are key molecular events of EMT. Transcription factors, such as Snail and Twist, bind to consensus E-box sequences in the E-cadherin gene promoter and down-regulate E-cadherin transcription [
46,
47]. The association between EMT and CSC has been reported previously. Several studies have provided evidence showing that CSCs express EMT markers and that induction of EMT could convert epithelial cells into breast CSCs [
27‐
30]. This demonstrates the essential role of EMT in CSCs acquiring invasive and metastatic phenotypes. We have proven our hypothesis that EMT is involved in the origin of migratory CSCs in colon cancer, using real-time RT-PCR to determine EMT-related gene expression. Pang
et al. reported that EMT-like attributes contribute to the invasive phenotype and metastatic capacity of the migratory subpopulation in CRCs [
41]. This is in line with our findings that the corresponding alteration in mRNA expression levels of EMT-related genes and higher migratory and invasive capacities have been observed in CD133
+CXCR4
+ cancer cells. Furthermore, we found that treatment with SDF-1 could further induce the occurrence of EMT in CD133
+CXCR4
+ cancer cells. The above data indicate that the CD133
+CXCR4
+ subpopulation contributes to liver metastasis of colorectal cancer
via EMT.
Consistent with our findings, Esther and colleagues demonstrated that transforming growth factor-β (TGF-β) induced the EMT process and de-differentiation in Fao rat hepatoma cells. This process coincided with upregulated CXCR4 expression and also sensitization of these cells to respond to SDF-1, which mediated migration [
48]. Similar results were observed in oral squamous cell carcinoma [
26,
49]. However, the reason cancer cells that have undergone EMT have a higher expression of CXCR4 is far from clear. Exploring the origin of migratory CSCs warrants further research and requires integration of current tumor initiation and progression concepts, including CSC, EMT, accumulation of genetic alterations and the tumor environment as driving forces [
13]. A deeper understanding of these factors could provide further insights into tumor biology.
The CSC hypothesis suggests that CSCs are a minority population that has the potential to self-renew, differentiate and regenerate a phenocopy of the original tumor. They would seem the most probable candidates that are resistant to chemotherapy, and they have been investigated previously [
3,
5,
50‐
52]. Novel treatments targeting CSCs may result in the complete eradication of tumor growth, and furthermore, based on the migratory CSC theory, if treatment targeting migratory CSCs can be developed, it might be possible to prevent tumor metastasis. We hypothesized that blockade of the SDF-1/CXCR4 axis might suppress colon cancer metastasis to the liver, with the knowledge that the liver secretes high amounts of SDF-1 [
53]. This is also in line with the theory that organs producing SDF-1 attract CXCR4
+ tumor cells and form metastatic tumors analogous to the directed homing of leukocytes. In our study, a nude mouse hepatic metastasis model was employed, and the results indicated that chemical inhibition of CXCR4 with AMD3100 could inhibit colon cancer metastasis to the liver. The anti-metastasis effect caused by the blockade of the SDF-1/CXCR4 axis is supported by another report [
54]. This finding provides important clues for the development of a targeted therapy in the treatment of CRC.
To validate the above findings in
in vitro experimental and in animal models, we carried out a prospective study to investigate whether CD133
+CXCR4
+ cancer cell content was associated with disease progression and prognosis. Statistical analysis showed that high CD133
+CXCR4
+ cell content is associated with poor 2-year survival of colorectal cancer patients. The clinical data provide evidence to support our hypothesis that double positive cancer cells might be involved in the metastatic process. Our data showed that cancer located in the rectum was associated with a high content of CD133
+CXCR4
+ cancer cell compared with colon cancer. This might be due to higher CXCR4 expression in rectal cancer than in colon cancer[
20], suggesting that the percentage of CD133
+CXCR4
+ cancer cells in future studies should be investigated separately in colon and rectal cancer rather than in a mixed way.
Acknowledgements
This project was supported by Key project of the National Natural Science Foundation of China(Grant NO: 81030041); Key Basic Research Project of China (Grant NO: 2010CB945600, 2011CB966200); National Natural Science Foundation of China (Grant NO: 30870974, 30801347, 30901722, 31171321, 81000970, 81101622, 30973433, 30801094); Special Funds for National key Sci-Tech Special Project of China (Grant NO: 2008ZX10002-019, 2008ZX10002-025); Shanghai Science and Technology Committee (Grant NO: 10ZR1439600, 11ZR1449500, 10411963100, 10ZR1439900, 11nm0504700, 09QA1407200, 07zR14143, 2008B009); Shanghai Municipal Health Bureau(Grant NO: XYQ2011044); Science Fund for Creative Research Groups, NSFC, China(Grant NO: 30921006); Stem Cell and Medicine Research Center's Innovation Research Program(NO:SCMRC1102) and 2011 Ph.D. Graduate Innovation Fund Project(NO:27).
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
Participated in the conception and design of the study and the critical revision of the manuscript for important intellectual content: LZ, LW, SZ, YJ, ZH. Performed the data collection and analysis: ST, HW, YW, RL, YY, XZ, XX, EY, YR, HL. Interpreted the data and produced the draft of the manuscript: SZ, ZH, ST, TL, HW, YW, RL, YY, XZ, XX, EY, YR, HL. Obtained funding for the study: LZ, LW. All authors read and approved the final version of the manuscript.