Ethical approval for the present study was obtained from the ethics committee of Nanfang Hospital, Southern Medical University (Guangzhou, China). Written informed consent was obtained from all patients and all clinical investigations were conducted according to the principles expressed in the Declaration of Helsinki. All animal experimental procedures were conducted in strict accordance with the recommendations of the relevant national and international guidelines for the care and use of laboratory animals (35). The protocol was approved by the Committee on the Ethics of Animal Experiments of the Department of Laboratory Animal Science, Southern Medical University. All efforts were made to minimize suffering.

CRC tissue fragments were obtained from 74 patients undergoing resection of colon adenocarcinoma between November 2007 and December 2010, including 45 males and 29 females. The patients were between 28–84 years old (mean age, 56 years), containing 32 cases located on the rectum, 14 cases on the sigmoid colon, 15 cases on the right colon, 9 cases on the left colon and 4 cases on the transverse colon. Written informed consent was obtained for tumor tissue collection and its use in pathological and immunohistochemical studies according to the ethical standards of the institutional review boards for human research at Nanfang Hospital, Southern Medical University and Shanxi Provincial Cancer Hospital (Taiyuan, China). Histological grade, tumor stage and tumor size were classified according to the standards of the International Union Against Cancer Tumor-Node-Metastasis classification system (36). Tissue samples were fixed by 10% formalin for 24 h at 37°C, embedded in paraffin, sliced at 5 µm and stained with 0.2% hematoxylin for 3 min and 0.5-1% eosin for 1 min at 37°C. The slices were then subjected to microscopic pathological diagnosis by two pathologists (Department of Pathology, Nanfang Hospital, Southern Medical University), who were unaware of the clinical observations.

Immunohistochemical staining for CD133 staining was performed using mouse anti-CD133 antibody (cat. no. AC133; Miltenyi Biotec GmbH, Bergisch Gladbach, Germany), as described previously (37). The slides were incubated with 1% bovine serum albumin (cat. no. B2064; Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) for 1 h at 4°C to block non-specific binding. Subsequently, each slide was incubated overnight at 4°C with anti-CD133 antibody at a dilution of 1:100, washed 3 times with PBS and then incubated with 50 µl horseradish peroxidase-labeled goat anti-mouse IgG (cat. no. TA130004; 1:100; Origene Technologies, Inc., Beijing, China) for 2 h at room temperature. The results of immunohistochemical CD133 expression were interpreted by Fromowitz semi-quantitative method (38). The intensity of staining and the percentage of staining positive cells was recorded in 10 high power fields, and the combination of the two methods were used to score, as follows: 0, negative (−); 1–2, positive (+/++); 3, strong positive (+++). A BX40 Optical microscope (magnification, ×200; Olympus Corporation, Tokyo, Japan) and MODEL5410 Research Grade Inverted Universal Material Fluorescent Microscope (Carl Zeiss AG, Oberkochen, Germany) were used.

The human CRC cell lines SW480, SW620, LoVo, HCT116, HT29 and Colo205 (American Type Culture Collection, Manassas, VA, USA) were cultured in RPMI-1640 (Hyclone; GE Healthcare Life Sciences, Logan, UT, USA) supplemented with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA), 100 U/ml penicillin G, and 100 µg/ml streptomycin (Gibco; Thermo Fisher Scientific, Inc.). All cells were maintained at 37°C in a humidified 5% CO₂ incubator.

CD133⁺ SW480 cells were selected by MACS using a CD133 Isolation kit (Miltenyi Biotec GmbH). This was conducted by centrifuging the cell suspension (300 × g for 10 min at room temperature), aspirating the supernatant completely and resuspending the cell pellet in 60 µl sorting buffer (containing 0.01 M PBS and 2 mM ethylenediamine tetraacetic acid) per 1×10⁷ total cells. Subsequently, 20 µl FcR Blocking reagent and 20 µl CD133 MicroBeads-Tumor Tissue (Miltenyi Biotec GmbH) was added per 1×10⁷ total cells. Following this, the cells were mixed well and incubated for 15 min in a refrigerator (2–8°C) under slow, continuous rotation using a MACSmix Tube Rotator. The cells were washed twice with sorting buffer and the supernatant was aspirated completely. Following this, magnetic separation was conducted. The cell suspension is loaded onto a MACS Column, which is placed in the magnetic field of a MACS Separator. The magnetically labeled CD133⁺ cells are retained within the column. The unlabeled cells run through; this cell fraction is thus depleted of CD133⁺ cells. After removing the column from the magnetic field, the magnetically retained CD133⁺ cells can be eluted as the positively selected cell fraction. To increase the purity, the positively selected cell fraction containing the CD133⁺ cells must be separated over a second column. MACS Column or MACS Separator were selected for use according to the number of total cells and the number of CD133⁺ cells. The enriched population of CD133⁺ cells was then suspended in serum-free culture system/medium, which contained Dulbecco's modified Eagle's medium/F12 (cat. no. SH30023.01; Hyclone; GE Healthcare Life Sciences), B27 (dilution 1:50; cat. no. 17504-044; Invitrogen; Thermo Fisher Scientific, Inc.), F12 EGF (20 ng/ml; cat. no. E5306; Sigma-Aldrich; Merck KGaA) and bFGF (10 ng/ml; cat. no. 233-FB-025; R&D Systems, Inc., Minneapolis, MN, USA).

CD133 protein expression on the cell membrane was analyzed by flow cytometry in the 6 human CRC cell lines, in the logarithmic growth phase. Cells were digested with 0.25% trypsin and 1 mM EDTA. Following incubation with 10 µl FcR blocking reagent for 10 min at 4°C (dilution, 1:50; cat. no. 130-092-575; Miltenyi Biotec GmbH), to reduce unwanted binding of antibody to Fc receptor-expressing cell, tumor cells were washed twice by washing buffer (containing 0.01 M PBS, 0.5% bovine serum albumin and 2 mM ethylenediamine tetraacetic acid) and resuspended at a concentration of 1×10⁶ cells/ml. Subsequently, CD133/2-PE antibody (dilution, 1:50; cat. no. 130-112-315; Miltenyi Biotec GmbH) was added, mixed well and incubated for 10 min in the dark at 4°C. The CD133/2-PE antibody dilution for labeling of cells and subsequent analysis by flow cytometry was 1:50 for 1×10⁶ cells/100 µl. Cell membrane CD133 protein expression was detected with a flow cytometer. Anti-IgG1 antibody (4°C for 10 min; dilution 1:50; cat. no. 130-117-098; Miltenyi Biotec GmbH) was utilized as a negative control. Data were acquired and analyzed using the FlowJo 10.0 software (FlowJo LLC, Ashland, OR, USA).

CD133⁺ and CD133⁻ cells were separated by MACS using a CD133 Isolation kit, according to the manufacturer's protocols. After adding 10 µl FcR (dilution, 1:50) blocking reagent per 1×10⁸ total cells at 37°C for 15 min, cells were immediately incubated with CD133/1-PE antibody (dilution 1:50; cat. no. 130-110-962; Miltenyi Biotec GmbH) directly labeled with supermagnetic microbeads for 30 min at 4-8°C. The cells were then washed by adding 1–2 ml separation buffer and centrifuged at 300 × g for 10 min at 4–8 °C. The pellet was resuspended in 500 µl separation buffer and processed to final separation using a Mini-MACs Separator Column (Miltenyi Biotec GmbH). Positively-labeled CD133⁺ cells were flushed with 2 ml separation buffer (containing 0.01 M PBS and 2 mM ethylenediamine tetraacetic acid) using a plunger fitted to the column, centrifuged for 5 min at 300 × g at 4–8 °C, and washed with washing buffer. Cell numbers and viability were determined using a hemocytometer with the standard trypan blue method (0.4% trypan blue; Invitrogen; Thermo Fisher Scientific, Inc.) (39). To assess the sorting, selected cells were incubated with CD133/2-PE antibody and analyzed using flow cytometry as aforementioned.

Non-obese diabetic/severe combined immune deficiency (NOD/SCID) mice weighing 20–25 g (n=5; male; 10–12 weeks old) were purchased from the Laboratory Animal Center at Southern Medical University (Guangzhou, China) and used to evaluate the tumorigenesis of CD133⁺ and CD133⁻ SW480 cells sorted by MACS in vivo. Mice were allowed to acclimatize for 1 week after arrival. All animals were maintained in a sterile environment with a daily 12/12 h light/dark cycle, the ambient temperature was maintained at 20–26 °C and the relative humidity ass 40–70 %. All animals were fed with relatively stable full-price nutritional feed and sterilized drinking water, which was added and replaced 3–4 times a week. After resuspension in 100 µl PBS, CD133⁺ and CD133⁻ SW480 cells (1×10⁴ cells/ml) were mixed with 100 µl Matrigel (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) and injected subcutaneously into the left and right abdominal walls, respectively (n=5 mice). Tumor volume was observed weekly for 5 weeks. All animal experiments were conducted in strict accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals (35).

Total RNA was extracted from CD133⁺ and CD133⁻ cells sorted from SW480 cells by MACS using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.). The quality and quantity of total RNA were estimated using a NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific, Inc.). An aliquot of 500 ng total RNA was reverse transcribed into cDNA using a PrimeScript^® RT reagent kit (cat. no. DRR037A; Takara Bio, Inc., Otsu, Japan).

Microarray analysis was performed according to the standard protocol (40) provided by CapitalBio Technology, Inc. (Beijing, China). cDNA labeled with a fluorescent dye (Cy5 and Cy3-dCTP) was prepared with the Eberwine's linear RNA amplification method (41) and subsequent enzymatic reaction. As a measure of technical replication, dye-swap experiments were performed on each of the biological samples. Arrays were scanned with a LuxScan 10KA dual channel laser scanner (CapitalBio Technology, Inc.) and the images obtained were then analyzed using LuxScan™ 3.0 software (CapitalBio Technology, Inc.). Space-and intensity-dependent normalization based on the LOWESS program was employed (42,43). Statistical analysis of Pathway (https://www.kegg.jp/) and Gene Ontology (GO) terms was performed in the Molecular Annotation System V4.0 (http://www.capitalbio.com/lifescience/informationsystems/2572.shtml), a web client program for interactive navigation within the knowledge bases, by selecting the differentially expressed genes. Data analysis utilized Pathway and GO saliency statistics. The results demonstrated the differentially expressed pathways that were involved in significant Pathway and GO terms.

Total RNA was extracted from CD133⁺ and CD133⁻ SW480 cells sorted by MACS using TRIzol reagent. cDNA synthesis was performed using a reverse transcription kit (Invitrogen; Thermo Fisher Scientific, Inc.), according to the manufacturer's protocol. RT-qPCR was performed using a SYBR^® Premix Ex Taq™ II kit (cat. no. DRR081A; Takara Bio, Inc.) with an ABI 7500 RT-qPCR System (Applied Biosystems; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocols. The reaction condition was as follows: 95°C for 2 min, 95°C for 15 sec and 60°C for 30 sec for a total of 40 cycles. The primers used [including p53, AKT1, transforming growth factor β receptor 2 (TGFBR2), ATP binding cassette subfamily B member 6 (ABCB6), cyclin dependent kinase 1 (CDC2), checkpoint kinase 1 (CHEK1), cyclin dependent kinase inhibitor 1A (CDKN1A) and cyclin B1 (CCNB1)] are listed in Table I. Expression levels of the GAPDH were used for normalization and quantification of gene expression 2^−ΔΔCq levels (44).

Total protein was purified from CD133⁺ and CD133⁻ SW480 cells sorted by MACS. Following adding radioimmunoprecipitation lysis buffer (containing 0.01 M PBS and 2 mM ethylenediamine tetraacetic acid), samples were centrifuged at 12,000 × g for 20 min at 4°C. Protein concentrations in the supernatants were measured using a bicinchoninic acid assay kit (Pierce; Thermo Fisher Scientific, Inc.), according to the manufacturer's protocols. Western blotting was performed by electrophoresis of 30 µg protein on a 10% SDS-PAGE followed by transfer to a polyvinylidene fluoride membrane (EMD Millipore, Billerica, MA, USA). The membrane was blocked by incubation in Tris-buffered saline (TBS) containing 0.1% Tween (TBS-T) and 5% nonfat dry milk for 1 h at room temperature, and washed three times with TBS-T. The membranes were incubated with antibodies against extracellular signal-regulated kinase (ERK)1/2, c-Jun N-terminal kinase (JNK), p38, phosphor (p)-ERK1/2, p-p38, p-JNK (all from the same kit; cat. no. 9926), p-CDC2, p-p53, p-CHEK1, p-CHEK2, p-retinoblastoma 1 (Rb1) and p-Rb2 (all from the same kit; cat. no. 9917) (all dilutions, 1:1,000; all from Cell Signaling Technology, Inc., Danvers, MA, USA), and β-actin and β-tubulin (all dilutions, 1:200; all from Santa Cruz Biotechnology, Inc., Dallas, TX, USA) overnight at 4°C. The blots were washed three times with TBS-T and incubated with the appropriate horseradish peroxidase-conjugated anti-rabbit IgG secondary antibodies (1:2,000; Origene Technologies, Inc.) for 1 h at room temperature. Following incubation with the secondary antibody, the membranes were washed again three times with TBS-T. The blots were developed by enhanced chemiluminescence (ECL) using a chemiluminescence ECL Western Blotting Detection kit reagent (BioVision, Inc., Milpitas, CA, USA) and detected using the LAS-3000 imaging system.

Statistical analysis was performed using Microsoft Office Excel 2007 (Microsoft Corporation, Redmond, WA, USA) and SPSS 13.0 statistical software (SPSS, Inc., Chicago, IL, USA). The frequencies of CD133⁺ expression under different features of patients were provided. In the association analysis of CD133 expression with features of patients, the CD133⁺ expression between two groups were compared using Mann-Whitney U-test and between multiple groups were compared using Kruskal-Wallis H-test. If the multiple groups comparison was significant, a post-hoc test using Mann-Whitney U-test with Bonferroni's P-value adjust method was performed. The expression of associated genes and proteins in CD133⁺ (+/++/+++) and CD133⁻ cell groups were compared using two independent samples Student's t-tests. Two-sided P<0.05 was considered to indicate a statistically significant difference.

Previous studies on the biological importance of CSCs identified human CD133 as a representative marker for CSCs (5,9,24,25). Although previous studies reported that CD133 expression was associated with lymph node metastasis (45–47), there is limited knowledge regarding other clinical outcomes associated with high-CD133-expressing CRCs. Therefore, CD133 expression levels in CRC tissues from 74 patients were examined by immunohistochemistry. CD133 protein was overexpressed in 25 tissues (33.8%) and exhibited reduced expression in para-cancerous tissues (Fig. 1A-F). Consistent with previous results, CD133 overexpression was not significantly associated with age, sex, location or classification, but CD133 expression was significantly increased in patients with poorly differentiated tumors (P=0.021) and distant metastasis (P=0.007). The patients were further divided into those with lower stage (no lymph node involvement or distant metastasis) and higher stage (presence of lymph node and/or distant metastasis) tumors, and it was demonstrated that CD133 overexpression was also significantly associated with tumor stage (P=0.018). These results indicated that CD133 overexpression is associated with disease progression in CRC (Table II).

The CSCs obtained from CRC cell lines were characterized by selection in serum-free culture system. Previous studies demonstrated that CD133 is an appropriate marker for detecting colon CSCs in primary tumor and cancer cell lines (32–37). Therefore, the percentage of CD133⁺ cells was assessed using flow cytometry. The proportions of CD133⁺ cells were increased in SW480, SW620 and LoVo cell lines, compared with the other three cell lines examined, with the highest proportion (9.16%) in SW480 cells (Fig. 2A). CD133⁺ and CD133⁻ cells were then purified from SW480 cells by MACS using a CD133 Isolation kit. The purity of CD133⁺ cells after the first separation was ~30%, but this increased substantially to >80% following the second separation (Fig. 2B). The enriched population of CD133⁺ cells was then suspended in serum-free culture system and grown as single cells, compared with CD133⁻ cells (Fig. 2C and D). Subsequently the molecular features of colon CSCs were investigated by examining the expression of genes involved in stem cell-associated pathways by RT-qPCR in CD133⁺ and CD133⁻ populations.

CD133⁺ cells obtained from CRC cell lines have been reported to be able to proliferate as sphere-like cellular aggregates in serum-free stem cell medium (32). Therefore, the enriched population of CD133⁺ cells was suspended in serum-free culture system and grown as single cells. In the first 2–3 days of culture, the majority of the cells began to proliferate, while a number of the cells died, and cell debris was observed at the bottom of the Petri dishes. On the 5th day, the cells continued to grow and formed tumor spheres with bright refraction and round shapes, but cell shapes or cell-cell boundaries could not be accurately distinguished. Once the size of the spheres reached 75–100 µm, they could be passaged. As the passage number increased, it was observed that the size of the tumor spheres remained relatively unchanged while their shape became more spherical (Fig. 2E-G). In the serum-free culture system, CRC stem cells were gradually enriched. In order to make the visual field clearer, enriched CRC stem cells were inoculated into a 96-well plate and to ensure that there was a single sphere in each field of view. The spheres were then imaged and their entire growth process was observed. CD133⁺ cells re-cultured in serum-containing medium adhered to the dish walls and formed a monolayer. These results demonstrated that CD133⁺ SW480 cells possessed stem/progenitor cell-like properties, including differentiation capacity.

Additionally, whether CD133⁺ colon cells were more tumorigenic than their CD133⁻ counterparts in vivo was also investigated. Each NOD/SCID mouse was injected subcutaneously in the left and right abdominal walls with 1×10⁴ CD133⁺ and CD133⁻ cells, respectively. Tumors were detected in the left abdominal wall of NOD/SCID mice 2 weeks after subcutaneous injection of CD133⁺ cells, but no tumors in the right wall (Fig. 3A and B). The tumors were >1 cm³ after 3–4 weeks. Transplanted tumors were confirmed as colon cancer by hematoxylin and eosin staining (Fig. 3C and D). These results demonstrated that CD133⁺ cells possessed increased tumorigenic capacity, compared with CD133⁻ cells.

Potential genes associated with the distinct cellular behaviors of the two cell populations were identified by generating gene expression profiles of CD133⁺ and CD133⁻ SW480 cells using a 22K Human Genome Array. A total of 21,522 genes were detected and 414 differentially expressed genes were observed, of which 185 were upregulated and 229 were downregulated in CD133⁺ cells (Fig. 4A). These genes included BMI1, which was previously reported to be involved in regulating tumor growth and the epithelial-mesenchymal transition process (48). A number of ABC transporters were also upregulated in CD133⁺ cells, including ABC subfamily B member 6 (ABCB6), ABCC2, ABCA7 and ABCF3, which may be associated with drug resistance in colorectal CSCs. Furthermore, the expression levels of anti-apoptosis genes, including interferon regulatory factor-3 (IRF-3), IRF-7 and death associated protein kinase 3, were increased in CD133⁺ cells, compared with CD133⁻ cells. Therefore, it was considered that the anti-apoptotic effect of colorectal CSCs was exerted by activating or inhibiting particular cell signaling pathways. These genes may be relevant to the malignant behavior of colorectal CSCs.

The differentially expressed genes associated with biological processes in CD133⁺ and CD133⁻ cell samples accounted for 55.6%, including transcriptional regulation, cell cycle, mitosis, transcription, cell division and protein amino acid phosphorylation. The differentially expressed genes associated with cell composition accounted for 16.1%, including nuclear, cytoplasm, cytosol, cell membrane, cell membrane integrity, cytoplasmic membrane and mitochondrial composition. The differentially expressed genes associated with molecular function accounted for 28.3%, including protein binding, nucleic acid binding, transferase activity, ATP binding, zinc ion binding, GTP binding and transcription factor activity. GO analysis demonstrated that the differentially expressed genes were primarily distributed in the cellular process (14.7%), the physiological process (13.2%), the metabolism (7.10%), the biological regulation (6.80%), and the regulation of biological processes (6.22%), cells (5.38%), binding (4.51%), developmental processes (4.20%), catalytic activity (4.20%) and other functions.

Gene pathway analysis revealed that the differentially expressed genes were distributed in multiple pathways, including the cell cycle, p53, tight junction, glioma, calcium, insulin, proteasome, gonadotropin-releasing hormone and MAPK signaling pathways (Table III). A total of 6 differentially expressed genes associated with the MAPK signaling pathway and 6 differentially expressed genes associated with the p53 signaling pathway were identified. Therefore, it was hypothesized that CD133 regulation of CRC occurrence and development may be associated with the MAPK and p53 signaling pathways. To examine this hypothesis, the expression levels of representative genes were examined by RT-qPCR to verify the results of the cDNA gene chip analysis. The mRNA levels of p53, AKT1, TGFBR2 and ABCB6 were significantly increased in CD133⁺ cells, compared with CD133⁻ cells (P<0.05; Fig. 4B), while the mRNA levels of CDC2, CHEK1, CDKN1A and CCNB1 were significantly downregulated in CD133⁺ cells (P<0.05; Fig. 4B). These results were consistent with the gene chip results (with the exception of CHEK1 mRNA), indicating that the microarray results were generally reliable. Subsequently, the protein expression levels of these key molecules were examined by western blotting. There were no significant difference in ERK1/2, p38, JNK and p-JNK protein expression levels between CD133⁺ and CD133⁻ cells. However, expression levels of p-ERK1/2 and p-p38 proteins were increased in CD133⁺ cells, compared with CD133⁻ cells (Fig. 4C). Furthermore, all the phosphorylated proteins involved in the p53 signaling pathway, including p-p53, p-CDC2, p-CHEK1 (Ser345), p-CHEK2 (Tyr68), p-Rb (Ser795) and p-Rb (Ser807/811), had reduced expression in CD133⁺ cells, compared with CD133⁻ cells (Fig. 4D and E). These results indicated that CD133 may assist CRCs to invade and metastasize by activating the MAPK signaling pathway and inhibiting the p53 signaling pathway.