Abstract
Background: Human cancer is characterized by high heterogeneity in gene expression, varieties of differentiation phenotypes and tumor-host interrelations. Growing evidence suggests that tumor-initiating, or cancer stem cells (CSCs), may also represent a heterogeneous population. The present study was undertaken to isolate and characterize the different phenotypic subpopulations of metastatic colon cancer and to develop a working colon CSC model for obtaining highly tumorigenic and clonogenic cells in sufficient numbers. Materials and Methods: Different phenotypic cell subpopulations were isolated based on differential levels and patterns of expression of several stemness markers, including CD133, CD44, CD166 and CD49b. Stemness properties of isolated cells were tested by analysis of their ability to form floating colonospheres in vitro, to induce tumors in NOD/SCID mice after transplantation at relatively low cell numbers, and to produce progenitors of different phenotypes. Results: The metastatic colon cancer HCT116 cell line, which expressed a majority of known CSC markers, closely resembling the patterns of expression in exfoliated peritoneal cells from several metastatic colon cancer patients, was selected as a reference material. Genome-wide microarray analysis (Affymetrix; DAVID) of CD133high CSC-enriched versus CSC-depleted cell populations revealed 4,351 differentially expressed genes with an overrepresentation of those responsible for apoptosis resistance, regulation of cell cycle, proliferation, stemness and developmental pathways. Simultaneous analysis of 84 stem cell- and metastasis-related genes with corresponding PCR arrays identified genes differentially expressed in several colon CSC phenotypic populations versus bulk tumor cells, and in relation to each other. It was found that colonospheres induced by tumorigenic cells with the highest expression of CD133 and those which were induced by CD133/CD44-negative cells possessed profoundly different stem cell-related gene expression profiles. Conclusion: The proposed approaches allow for reliable isolation and propagation of highly tumorigenic and clonogenic cells of different phenotypes. Genomic analysis of several candidate CSC phenotypic populations may contribute to the identification of novel targets for colon cancer stem cell-targeted drug development and treatment.
The limited effectiveness of standard anticancer therapies has been attributed to the existence of rare, highly drug resistant cells within tumors, operationally called cancer stem cells (CSCs) (1). This concept was firmly supported by the recent isolation and characterization of tumorigenic cells from all major human hematopoietic and solid tumors, which led to the important conclusion that not every cancer cell, but only CSCs are responsible for tumor initiation, development and response to treatment (2, 3). CSCs share some basic features and signal transduction pathways, such as Wnt, Shh, Notch, Bmi-1 and others with normal stem cells (4, 5). Thus, it is widely accepted that CD133 [known as prominin-1 or AC133, a pentaspan glycoprotein (6) originally classified as a marker of primitive haematopoietic and neural stem cells] is also a marker of organ-specific adult stem cells and CSCs, and several types of tumorigenic cells, including brain tumors (7-9), kidney (10), hepatocellular (11, 12), colon (13, 14), pancreatic (15, 16) and prostate (17) carcinomas were isolated using this marker. These studies have demonstrated that the minor CD133+ cell population possesses much higher tumorigenic and clonogenic potential compared to their CD133- counterparts or unsorted cells, and that the CD133+ cells are highly resistant to chemo- (18-19) and radiotherapy (20).
Human cancer is characterized by high heterogeneity in gene expression, varieties of differentiation phenotypes and interactions between cells and the extracellular matrix, which tightly regulate tumor features, growth rates, drug sensitivity, etc. There is accumulating data that CSCs may also represent heterogeneous phenotypic populations. Thus, several distinct CSC populations were found in breast tumors: CD133+ and CD44+/CD24- (21); and CD24+/CD29+ or CD24+/CD49f+ (22). A subpopulation of migrating CD133+/CXCR4+ CSCs was identified in pancreatic tumors (15). Colon cancer specimens possess high variability in the patterns and levels of expression of some stem cell markers, such as CD133 (23, 24), which is in line with data showing that the percentage of CD133+ cells correlates with tumor aggressiveness and clinical outcome (9, 25-27). Although the isolation of CD133+ CSC is based on the assumption that differentiated cells lose their CD133-positivity, several reports suggested that CD133 is rather widely expressed in some epithelial tissues, including kidney cancer and lactiferous ducts of mammary glands (28), mature epithelium of pancreatic ducts (29) and primary colon cancer cells (24). Since not all solid colon tumors express CD133, the combined expression of EpCAMhigh/CD44+/ CD166+ was suggested as being more robust for the isolation of colon tumor-initiating cells (23). There is growing evidence that both CD133+ and CD133- populations are tumorigenic in some cases of advanced metastatic glioblastoma (30-32) and colon cancer (24), although the lack of CD133 by itself does not indicate an exact phenotype of this candidate CSC population. It is yet to be established what cancer cell subpopulations are responsible for initiation and progression of each tumor type, and what the nature of each type of tumor-initiating cell is. The present study was undertaken to isolate and characterize the different phenotypic subpopulations of metastatic colon cancer and to develop a working colon CSC model for obtaining highly tumorigenic and clonogenic cells in sufficient numbers. The metastatic colon cancer HCT116 cell line, which expresses a majority of known CSC markers, and possesses a constitutively larger proportion of several stemness markers was selected as a reference.
Materials and Methods
Isolation of CSCs from peritoneal wash of colon cancer patients. Patients with clinically advanced metastatic colon cancer subjected to surgical treatment were asked to participate in this study by providing peritoneal washing fluid and signing an informed consent in accordance with the local IRB requirements. At the beginning of surgery, 700 ml of sterile saline were instilled into the peritoneal cavity, gently agitated by rotation of the operative table and aspirated. Peritoneal washes were immediately delivered from the operating room and centrifuged at 300 xg for 6 min at 4°C. The cell pellet was reconstituted in PBS and gently loaded onto a layer of Histopaque-1077 gradient (Sigma-Aldrich, St. Louis, MO, USA) and centrifuged at 400 × g for 30 minutes at room temperature. RBC, dead cells and debris were removed from the bottom of the tube and live nucleated cells collected at the interface.
Cells. Standard HCT116 cells derived from a poorly-differentiated human colonic adenocarcinoma were grown under standard conditions (DMEM with 10% FCS, on uncoated flasks or dishes). Isolated CD133+ cell populations were cultured in serum-free medium supplemented with 20 ng/ml EGF, 10 ng/ml FGF-2 and 10-20 ng/ml LIF (optional), on collagen type I-coated dishes.
Magnetic Cell Sorting (MACS) and antibodies. Dissociated cells were centrifuged at 300 xg for 6 min at 4°C, rinsed with sterile MACS buffer and labeled with CD133 antibodies (Abs) directly or indirectly conjugated with ferromagnetic beads (Miltenyi Biotec, CA, USA) as recommended by the manufacturer. For up to 10 million cells, 350 μl MACS buffer, 100 μl of the blocking reagent and 50 μl of the biotinylated CD133 were added, mixed well and incubated for 15 min at 4°C. After washing twice with 7 ml MACS buffer cells were incubated for 20 min at 4°C with 100 μl Anti-Biotin in 400 μl MACS buffer. Washed labeled cells were resuspended in 500 μl of buffer and sorted with MACS devices.
Flow cytometry (FACS) and Abs. Cells were prepared as described above and labeled with one or several markers conjugated with different fluorescent dyes, including anti-human CD133/2-APC (clone 293C3; Miltenyi Biotec, CA, USA); CD166-PE (clone 105902; R&D Systems, MN, USA); CD44-FITC (clone F10-44-2), CD44-PE (clone F10-44-2; Invitrogen/Biosources, USA); CD44v6-FITC (clone 2F10; R&D Systems, USA), EpCAM-FITC (Biosources). Antibodies were diluted in buffer containing 5% BSA, 1mM EDTA and 15-20% blocking reagent (Miltenyi Biotec) to inhibit unspecific binding to non-target cells. After 15 min incubation at 4°C, stained cells were washed, resuspended in 500 μl of MACS buffer and sorted with multiparametric flow cytometry with BD FACSAria cell sorter (Becton Dickinson, CA, USA) under sterile conditions.
In vivo tumorigenicity. All experiments involving the use of animals were performed in accordance with SBU institutional animal welfare guidelines. NOD/SCID mice (Charles River Laboratories, Wilmington, MA, USA) were maintained under defined conditions at the SBU animal facility. Immediately after sorting, aliquots of the particular cell populations were counted and cell viability was determined by conventional trypan blue test. Cells were centrifuged at 300 xg for 6 min at 4°C, suspended in ice-cold 1:1 mixture of growth medium and Matrigel Matrix (BD Biosciences) and 40 μl were injected subcutaneously into the flanks of 8-week-old mice. The primary tumor sizes were measured with a caliper on a weekly basis. All mice were sacrificed by cervical dislocation after 6 weeks post-transplantation.
Analysis of mouse tumor xenografts. Tumors were minced with scissors under sterile conditions, rinsed with Hank's balanced salt solution (HBSS) and incubated for 2 hours at 37°C in serum-free RPMI-1640 medium supplemented with 200 units/ml Collagenases type II and type IV (Sigma-Aldrich, USA), 120 μg/l penicillin and 100 μg/mL streptomycin. Cells were further disaggregated by pipetting and serial filtration through cell dissociation sieves (size 40 and 80 meshes; Sigma-Aldrich). Contaminating erythrocytes were lysed by incubation in ammonium chloride hypotonic buffer for 5 min on ice. Single cell suspensions were sorted with BD FACSAria and subjected to further analysis.
Sphere formation assay. The clonogenic potential of different phenotypic populations was analyzed in 3D non-adherent culture conditions using 10% Matrigel matrix (BD Biosciences) in MSCBM (Lonza, MD, USA). Cells were counted, resuspended and plated on 6-well plates at a final count of 300 and 1×103 cells per well. One and two weeks after initiation, plates were inspected for colony (sphere) growth: the number of colonies within each well was counted under a microscope and representative fields were photographed.
Affymetrix GeneChip human genome U133A array. Five μg of good quality total RNA was added to a cDNA reaction using a T7 dT24 primer and SuperScript II RT (Invitrogen, USA). The cDNA was purified utilizing a phenol/chloroform extraction followed by an ethanol precipitation. An in vitro transcription (IVT) reaction was performed using the BioArray™ HighYield™ RNA Transcript Labeling Kit (Enzo Life Sciences, USA). The IVT reaction products were purified using RNAeasy purification kit (Qiagen, USA) with a DNAase treatment step and analyzed as recommended by the manufacturer (Affymetrix, USA). For each sample an Affymetrix U133 Plus 2.0 array was hybridized overnight at 45°C in an Affymetrix Hybridization Oven 640. The arrays were then washed, stained in an Affymetrix GeneChip® Fluidics Station 450, scanned in Affymetrix GeneChip® Scanner 3000 7G and the images were analyzed using the Affymetrix GCOS software version 1.4. The gene list with 2-fold cut-off was then further subdivided into functional categories with the bioinformatic analysis resource DAVID (Database for Annotation, Visualization and Integrated Discovery) of the Advanced Biomedical Computing Center (NCI, Frederick, http://david.abcc.ncifcrf.gov). Gene Pathway analysis was also analyzed by DAVID.
Results
Phenotypic analysis. It was found that a majority of colon cancer cells, either isolated from peritoneal washes of metastatic colon cancer patients (5 of 5 tested specimens; representative experiment is shown in Figure 1 A), or from HCT116 cell line (Figure 1B), were positive for CD133 and CD44, expressing low-to-moderate levels of these two markers and high levels of CD166 (Figure 1 C) and EpCAM (not shown). Both types of cells also possessed several distinctive minority cell populations with highest levels of expression of some stemness markers, including CD133high/CD44high, CD166high/CD44high, as well as minor populations of cells lacking CD133 (CD133-/CD44high and CD133-/CD44-; Figure 1); however, the expression levels in clinical specimens were variable. Since the HCT116 cell line possessed all these minority phenotypic populations at higher ratios than, for example SW480 and HT29 cells, it was selected as a major source for further studies.
Since CSCs usually represent a very minor population of cancer cells, several experimental approaches which promoted CSC self-renewal and allowed significant enrichment of CSC-associated phenotypic populations were tested. Thus, culturing cells at low density (30-50% confluent) before sorting led to a significant increase in the yield of MACS-CD133+ cells, ranging from 20.5% to 50.7% (38.5±12.7) compared to 4.9±1.1% if cells were cultured at 75-90% confluent culture. Since not all isolated CD133+ cells are putative stem cells (although enriched, it is a mixture of stem cells with their early progeny, which still retain CD133-positivity), it was important to determine conditions which could increase the proportion of putative CSCs. It was found that repeated sets of cell sorting followed by culturing at low cell density in serum-free MSCB medium led to a significant increase in the proportion of CD133high cells (Figure 1 D, E), which was retained to some extent after transplantation to NOD/SCID mice (Figure 1F).
Analysis of stem cell properties. The isolated phenotypic cell population was tested for stem cell properties, i.e. thir ability to form floating colonospheres in 3D culture systems (clonogenic potential), their ability to induce tumors in NOD/SCID mice after transplantation of a relatively low cell number (in vivo tumorigenicity) and their ability to produce different cells phenotypes (plasticity). To examine their in vitro clonogenic potential, an equal number of the freshly FACS purified cells (300 and 1×103 cells per well) were seeded on Ultra-Low Attachment 6-well plates (Corning, NY, USA) in 10% Matrigel matrix in serum-free stem cell medium. In 10 days, all produced spheres were counted and representative fields were micro-photographed. It was found that four phenotypic populations, including CD133high, CD133-, CD133high/CD44high and CD166high/CD44high, possessed much higher efficiency in forming floating colonospheres (Figure 2 A, B; only spheres induced by CD133high/CD44high and CD133- populations are shown), compared to unsorted bulk tumor cells, which produced predominantly flat, 2-dimensional colonies at non-adherent culturing conditions. The CD133- cells induced lower numbers, but larger sizes of colonospheres, with a higher number of produced cells (Figure 2 C, D). It was noted that long passaging of HCT cells as floating spheres, as well as repeated cell sorting, led to the shift of the major cell population toward higher levels of expression of CD133 and gradually increasing clonogenic capacity.
To compare phenotypic composition of colonospheres induced by different populations of colon cancer cells, all spheres of each phenotype were collected, dissociated in enzyme-free dissociation reagent (Chemicon Int.) and analyzed, sorted and counted with the BD FACSAria for further tests. Surprisingly, FACS analysis of dissociated spheres induced by CD133high, CD133-, CD133high/CD44high and CD166high/CD44high subpopulations revealed a strikingly similar pattern of expression of all tested cell surface CSC-related markers, with quite similar ratios of the minority phenotypic populations (Figure 2 A, B). All dissociated sphere cells possessed an additional distinctive minority population with high levels of expression CD49b (right column). These data indicate that the above-listed subpopulations of metastatic colon cancer cells have similar abilities to form colonospheres in 3D culture system and similar plasticity (ability to give rise to other phenotypes) under the described conditions.
In vivo tumorigenic potential. Cells of a particular phenotype were transplanted into NOD/SCID mice in quantities which were known to reliably and reproducibly induce mice tumor xenografts and, on the other hand, could also reveal possible differences in their tumorigenic potential. Thus, it was shown previously that 1,000 cells of several CSC phenotypes induced tumors in the majority, but not in all injected mice (23). Therefore, 1,500 cells of several candidate phenotypes were transplanted. It was found that MACS-CD133+ and FACS purified CD133high/CD44high, as well as cells which were repeatedly depleted of CD133+ population, uniformly induced tumors in NOD/SCID mice, whereas no tumors were developed after injection of 5×105 unsorted bulk HCT116 cells by 6 weeks after transplantation. Importantly, cells repeatedly sorted and cultured at low density possessed a much higher ability to form tumors compared to single-time sorted MACS-CD133+ cells, which means that most likely, they had an enriched proportion of putative tumorigenic cells, or CSCs. Thus, at 6 weeks after transplantation, 6 of 6 mice developed relatively small tumors after injection of x1 MACS-CD133+ cells, whereas repeatedly sorted cells induced significantly larger and more vascularized tumors in all mice (Figure 3A, B). Dissociated and MACS-CD133-sorted tumor xenograft cells also possessed significantly different sphere-forming capacities (Figure 3C), which means that an increased ratio of CD133high in transplanted cells was retained to some extent in mouse tumors. FACS analysis revealed that, similarly to floating colonospheres, dissociated mouse tumors induced by different subpopulations also possessed all the original studied phenotypes, including a majority population which was low-to-moderately positive for CD133 and CD44, and highly positive for CD166, as well as minority subpopulations expressing highest levels of CD133high, lack of it (CD133-), and these which were CD166high/CD44high (Figure 1D and Figure 4). Repeatedly sorted CD133high cells produced a significantly larger proportion of CD133- cells as early as 2 days after culturing of isolated cells in adherent conditions (5.8% versus 18%), and such an increased CD133- ratio was retained in mouse tumor xenografts induced by repeatedly sorted cells (Figure 4; right panel; lower right quadrant), as well as in floating colonospheres. As mentioned above, the CD133- subpopulation possessed profound clonogenic and tumorigenic properties.
Genome-wide gene expression profiling. Using metastatic colon cancer HCT116 cells as a reference material, genome-wide and several pathway-specific gene expression profiles of CSC-enriched CD133high cell population versus bulk tumor cells were analyzed. Expression of several stemness-related genes was also tested with real-time RT-PCR. Using the Affymetrix GeneChip Human Genome U133A assay it was found that freshly isolated CSC-enriched CD133high cells possessed 988 differentially expressed genes compared to their bulk counterparts (CD133high-depleted cells). Of these genes, CSCs showed significantly increased activity of 826 genes and reduced activity in 162 genes. The CD133high-induced 3D floating spheroids revealed 4,351 differentially expressed genes with more than 500 significantly up-regulated (3- to 120-fold) and 436 down-regulated (by 3- to 1,500-fold) genes. The majority of the most up-regulated genes were those related to: anti-apoptosis (APP , Bcl3/NFkappa B2 complex, BDNF, BIRC3, BIRC4, BTRC3, CBX4, CCAR1, CCPG1, CD74, DHCR24, FOXO3, HSPA1B, IGFBP3, IF16, NFKB1A, TBX3, TNFAIP3, TRIB3 and others), stemness (NOTCH; APP, MIB1; Wnt receptors TGFB1I1, CSNK1D), cell cycle/cell proliferation (FOSB, IL-8, CCNG2, IGFBP3, TGFBP1, MXD1, INSIG1, EHF, CD74, CDC25A, HSMPP8), transcription factors (ID2, ID2B, DENR, MXD1), DNA repair and many others. Overexpression of some stemness-related genes (NOTCH, Shh, Wnt, Oct4) was determined with RT-PCR (data are not shown). The most significantly down-regulated genes in HCT floating spheres were HLI4 (by 1,500-fold), which is responsible for heterophilic cell adhesion (motility, migration, metastasis); apoptosis-related cytochrome c, COX6A1 (by 300-fold) and BCL2L1, which regulates the release of cytochrome c from mitochondria; CXCL14 involved in cell-cell signaling (by 100-fold) also responsible for cell motility, migration and metastasis; apoptosis-related AP15, BAX, CASP2, CFL1, ENO1, FXR1, HSPD1, HSP90B1, FAS, Fas-binding (FBF1, NPM1), MVEGFA, RAD21, RHOB, SOCS2, VDAC1 and many others; cell cycle/cell proliferation (ras RHOB, CDV3, CDK8, NFYC); cell motility (TPM1, VIM, MACF1); genes involved in negative regulation of cell growth (DCBLD2, POSTN, CDH11); signal transduction (ATP binding: SPARC, MAP3K2, HSP90AB1) and heat-shock protein genes (HSP90B1, HSPD1), which are required for antigen presentation (which is usually lost in advanced cancer).
Affymetrix microarray data were analyzed with DAVID (NIH, Frederik). The gene list with 2-fold cut-off was then further subdivided into functional categories with this bioinformatic analysis resource. All genes were grouped based on functional similarity using novel clustering algorithms. A portion of the global view of genes is shown on the fuzzy heat map in Figure 5 A, B (two fields are described; red outline). A summary of several functional gene clusters is shown in Figure 5 C.
Pathway-specific gene expression profiling. To characterize more precisely the alterations in the pathway-specific gene expression, the SYBR Green-based RT Profiler PCR Array System (SA Biosciences) was used. Stem cell-related and metastasis-related gene expression profiles were analyzed in two candidate colon CSC phenotypes, CD133high and CD133- (which are also negative for CD44 and CD166) compared to their bulk counterparts, and in relation to each other. Both CD133high and CD133- colonospheres revealed dramatic up-regulation of the majority of stemness-related genes (51 and 63 out of 84 analyzed stem cell-associated genes, respectively) compared to the bulk HCT116 cells (Figure 6 A, B). It was determined that CD133-/CD44-/CD166- colon cancer cells have distinct growth characteristics and gene expression profiles. Thus, PCR array analysis revealed that the expressions of several stem cell-associated genes, including ABCG2, NOTCH1, NOTCH2, WNT1, SOX2, DLL1, DLL3, DTX1, DTX2, FGF3, FRAT, IGF1, CD44, CXCL12, BMP2, COL2A1, MYOD1, DVL1, CD3D, CD4, CCND2, CD8A, CDH2 and TERT, were more significantly up-regulated in colonospheres induced by CD133- cells compared to those induced by CD133high. On the other hand, CD133high-induced colonospheres also possessed a different set of genes, which were more significantly up-regulated compared to CD133- spheres, including ALPI, ASCL2, AXIN1, BMP1, BMP3, BTRC, CD30, CDC2, CTNNA1, EP300, FGF1, FGF4, FGFR1, FAG1, NEUROG2, PPARD, RB1, S100B, SOX1 and others. In contrast to the stem cell-related genes, the majority of the metastasis-related ones were slightly down-regulated in CSC-enriched populations versus bulk tumor cells; however, several genes were moderately up-regulated in CD133high - induced colonospheres (Figure 7), including CXCR4, VEGF and its receptor FLT4, CTSL, CXCL12, MMP3, MMP7, SYK, TIMP3, TRPM1 and TSHR.
Discussion
Currently, there are controversial data concerning the difference in the expression of the most commonly used CSC marker, CD133, as well as other markers, between colon tumorigenic cells, bulk tumor cells and normal colonic epithelium. It remains to be established which is more important: the presence/absence of a particular marker or levels of its expression. The objective of this study was to identify, isolate and characterize the phenotypic, genomic and some biological features of tumorigenic subpopulation(s) from the HCT116 metastatic colon cancer cell line, which can be used as an alternative, relatively standardized approach compared to the isolation of colon CSCs from clinical specimens. Presently, the isolation of CSCs from solid tumors has certain limitations, such as a lack of unique CSC markers, high heterogeneity of human cancer and high variability in gene expression, including the expression of stemness markers, resulting in difficulties in data interpretation and comparative analysis of the basic CSC features. Several different approaches are currently used for the isolation of colon CSCs, including CD133+ as a single marker (13, 14), combined expression of EpCAMhigh/CD44+CD166+ (23), lack of the CD133 expression (24), and side population (SP), based on the efflux of Hoechst 33342 (33, 34). The results clearly demonstrate that, in contrast to other cancer cell types, the majority of both exfoliated cells from peritoneal washes of metastatic colon cancer patients and cells from established metastatic colon cancer HCT116 cell line express low/moderate levels of CD133/CD44 and high levels of EpCAM/CD166. Therefore, positivity alone for these markers is not adequate as a criterion of stemness, at least for colon cancer, which is in line with the suggestion that quantitative differences in gene expression may be of high functional significance (35).
Several distinctive minority phenotypic populations have been identified, including CD133high, CD133-/CD166-/CD44-, CD133high/CD44high and CD166high/CD44high, which possessed significantly higher clonogenic and tumorigenic potentials compared to their bulk (although CD133-positive) counterparts. A high proportion of CD133+ cells in colon cancer HCT116 cells is not unique, and similar findings were recently reported for a liver cancer cell line (Huh-7), which exhibits as much as 50% CD133+ cells (11). There is accumulating data that the percentage of CD133+ CSCs correlates with tumor aggressiveness (8, 18, 25) and poor prognosis (19, 36), so it is not surprising that clinical specimens of colon tumors displayed high variability in CD133+ content, which was reported from 1.8 to 24.5% (13), 2.5% (14), and from entirely negative to predominantly CD133+ (23). Similar variability was reported for established colon cancer cell lines (37), showing that only 6 out of 12 lines expressed CD133 mRNA, and CD133 protein expression was detected in 5 out of those 6 cell lines. In contrast, exfoliated cancer cells isolated from the peritoneal cavity of patients with metastatic colon cancer represent a unique and naturally relatively enriched population of cells committed to induce metastasis after acquiring all the necessary mutations and migratory potential. Therefore, to determine possible differences in clonogenic and tumorigenic properties of several candidate phenotypic populations of colon cancer cells, the established metastatic colon cancer HCT116 cells were chosen as a reference line expressing a majority of commonly used stem cell markers, which closely mimics the basic phenotypic patterns observed in clinical specimens of the peritoneal washes.
MACS-CD133+ cells repeatedly sorted/cultured at low density in serum-free MSCBM revealed much higher in vivo tumorigenic potential and ability to form floating spheres, and the major population of repeatedly sorted cells generally expressed higher levels of CD133 as compared to the original unsorted cells, and a higher proportion of CD133high cells. It is suggested that this is the cell population with the highest levels of CD133 which represent putative colon tumorigenic cells, i.e. CSCs. This approach, however, did not affect the proportion of cells with combined highest expression of CD44/CD166. Since CD44high/CD166high cells did not co-express the highest levels of CD133, it can be suggested that they most likely represent two separate tumorigenic subpopulations. It has been determined that after 6-10 passages of colonospheres induced by CD133high cell population, a majority of cells acquired high clonogenic potential comparable the potential of CD133high cells. However, these data are in line with observations that malignant potential usually increases in consequent passages in vivo and in vitro due to the high rates of genetic and epigenetic changes. Thus, increased tumorigenicity and clonogenicity of cancer cells was reported after transplantation to mice (38). Moreover, secondary, tertiary and quaternary xenografts were shown to be increasingly tumorigenic, and doubling time within ten passages of colon cancer spheroid aggregates was longer than after 30 passages (39). It is known also that recurrent cancer is associated with more malignant phenotype with increased intrinsic or acquired drug resistance and high mortality rates (3).
It has been determined here that several distinctive phenotypic populations, including CD133high, CD133-/CD166-/CD44-, CD133high/CD44high, CD166high/CD44high and CD49bhigh have similar abilities to form colonospheres in 3D culture systems and similar plasticity (ability to give rise to other phenotypes) under the described conditions. These data are in line with another study on glioma (31), which demonstrated that both CD133+ and CD133- cells can produce both phenotypes. However, it contradicts another report on colon cancer (24) which suggested that the CD133- population can produce only CD133- cells. It remains to be established whether or not these candidate colon tumorigenic and clonogenic subpopulations have a common stem cell predecessor and what their exact functional role in primary and metastatic colon cancer is. Although these subpopulations have similar plasticities, dramatic genomic differences between several candidate phenotypic populations, as well as between CSC-enriched and bulk colon cancer cells, have been found using genome-wide and stem cell pathway-related gene expression analyses. Thus, up-regulation of the CXCR4 gene in CD133high cells compared to their bulk counterparts, which was identified in a subpopulation of migrating CSCs in pancreatic tumors (15), highlights the importance of acquiring a migrating phenotype by colon CSCs as a prerequisite for metastasis.
In conclusion, the present study suggests a working in vitro CSC model for obtaining a sufficient number of highly tumorigenic and clonogenic cells from a readily available metastatic colon cancer HCT116 cell line with further propagation of CSCs as colonospheres of different phenotypes in 3D culture. Significant up-regulation of many genes involved in distinct stem cell-related and developmental signaling pathways, in addition to the activation of anti-apoptotic pathways in metastatic colon CSCs, suggests that these features may contribute to sustained tumor growth and drug resistance. The combination of the genome-wide and pathway-specific analyses with microarray and PCR-array assays can provide extremely important and accurate information for the identification of potential targets for the development of effective anticancer therapies and obtaining gene signature patterns for each clinicopathological case. Further study is necessary to determine the biological significance of each phenotypic cell population.
Acknowledgements
We thank Todd Rueb for extraordinary technical support with flow cytometry and cell sorting experiments.
This work was supported by the NYSTAR-designated Center for Advanced Sensor Technology, NY LILAC foundation, and SBU SOM.
Footnotes
- Received November 27, 2008.
- Accepted December 24, 2008.
- Copyright© 2009 International Institute of Anticaner Research (Dr. John G. Delinassios), All rights reserved