It is known that many tumors are composed of heterogeneous populations of cancer cells with different phenotype and characteristics. The presence of cancer stem cells (CSCs) in tumors can be a source of such heterogeneity because they have the ability to divide asymmetrically and generate more differentiated cancer clones. CSCs represent potential targets for future anticancer therapies, as their capability to serve as reservoirs of cells that cause tumor recurrence has been demonstrated in various types of cancers including prostate and colon cancers. The exact cellular origin of CSCs needs to be fully elucidated. For example, in the prostate, it has been suggested that terminally differentiated prostate luminal cells give rise to prostate cancer (1), as most prostate cancer cells express luminal cell markers but lack basal cell markers (2). However, other studies suggested that prostate cancer could be induced from the intermediate progenitors or basal cells (3, 4). The consequence of such discrepancies is that no commonly accepted strategy exists to define and isolate normal stem cells (SCs) and CSCs from prostate cancer tissue. Surface markers such as CD44+/α2β1hi/CD133+ (5, 6) and CD49f+/Trop2+ (7) have been demonstrated to define normal human prostate SCs. Normal mouse prostate SCs are defined by the expression of Lin−/Sca-1+/CD133+/CD44+/CD117+ (8) or Lin−/Sca-1+/CD49f+/Trop2+ (7). In general, the existence of multiple subpopulations of CSCs with similarities to normal SCs corresponds with the high phenotypic heterogeneity within tumors.
Analyzing the biological characteristics and response of cell populations expressing CSC markers is challenging. The rare frequency of CSCs and their plasticity make it difficult to use common in vitro techniques. The gold standard assay for analyzing the self-renewal and lineage capacity of CSCs is serial transplantation in animal models. However, this approach is relatively time-consuming, and it is difficult to use in high-throughput screening to identify and validate drugs that target CSCs (9). The in vitro clonogenic assay is a commonly used method based on the ability of a single cell to grow into a colony (>50 cells). The clonogenic assay has been used in many studies to analyze the effects of various treatments (γ radiation, chemotherapeutics) in many cell types including SCs, cancer cells, and CSCs. Established protocols usually involve plating and cultivating 100–10,000 cells in semisolid medium in Petri dishes or 6- or 24-well microplates. The growth of cells in an anchorage-independent condition in semisolid medium does not necessarily correlate with the tumor-forming capability of the cells (10); however, this technique offers the advantage of the formation of physically detached colonies established from seeded cells in a particular well or dish. The clonogenic assay was originally introduced in 1956 by Puck and Markus (11) as a technique for assessing the clone-forming ability of single cells plated in vitro on irradiated feeder cells (12). Interestingly, it was recently demonstrated that the clone-forming ability of cells seeded on irradiated feeder cells could be significantly improved by Rho kinase (ROCK) inhibition (13). This cultivation condition induces normal and tumor epithelial cells from many tissues to proliferate indefinitely in vitro and represents an effective strategy for establishing cell lines (clones) from patient samples.
Flow cytometry using fluorescence-activated cell sorting (FACS) is a sensitive, specific, and robust approach for the phenotypic analysis and purification of particular cell populations including CSCs (9). In this study, we wanted to establish an automatic cell-cloning assay (ACCA) in a microwell format using an automatic cell deposition unit and a FACS system. Our results revealed that the novel ACCA is a straightforward approach for determining the clonogenic capacity of different subpopulations of cancer stem-like cells identified in both cell lines and patient samples.
Cells and Cell Culture
The mouse prostate cancer cell lines E2, E4, cE1, and cE2 (14) (a kind gift from Dr. Pradip Roy-Burman, University of Southern California, Los Angeles, CA) were cultivated in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen, Carlsbad, CA) supplemented with rhEGF (3 ng/mL, Sigma-Aldrich; St. Louis, MO, USA), insulin (1.25 μg/mL, Sigma-Aldrich), bovine pituitary extract (6.25 μg/mL, Hammond Cell Tech), penicillin (100 U/mL) and streptomycin (0.1 mg/mL; PAA, Pasching, Austria), and 5% fetal bovine serum (PAA). HCT-116 human colon adenocarcinoma cells (a kind gift from Dr. Bert Vogelstein, Johns Hopkins University, Baltimore, MD) were maintained in McCoy's 5A modified medium containing 1.5 mM L-glutamine (Sigma-Aldrich) supplemented with penicillin (100 U/mL) and streptomycin (0.1 mg/mL) (PAA), sodium bicarbonate (1.5 g/L, Serva; Heidelberg, Germany), and 10% heat-inactivated fetal bovine serum (PAA). A human fetal colon cell line (FHC) was obtained from ATCC (CRL-1831, LGC Standards, Poland). Cells were cultured in a 1:1 mixture of Ham's F12 medium and DMEM (Gibco, Invitrogen) supplemented with 25 mM HEPES, 10 ng/mL cholera toxin (Calbiochem, Merck, Germany), 0.005 mg/mL insulin, 0.005 mg/mL transferrin (Sigma-Aldrich), 100 ng/mL hydrocortisone (Sigma-Aldrich), and 10% heat-inactivated fetal bovine serum. Primary prostate cancer cells were isolated from patients undergoing radical prostatectomy (perinuclear tumor comprising ∼ 20% of the prostatic gland, Gleason score 4+3, N0, SV0). Post-surgery 18-gauge biopsy needle cores (Bard, Tempe, AZ) of prostate tissue were frozen in medium containing 10 μM Y-27632 dihydrochloride (ROCK inhibitor, Santa Cruz, Santa Cruz, CA) and processed as previously described (13). The study was approved by the ethics committee of the Faculty of Medicine and Dentistry, Palacky University, and informed consent was obtained from patient prior to surgery. Thawed biopsy specimens were digested by incubation for 5 h in dissociation solution containing 100 U/mL collagenase type I (Worthington, Lakewood, NJ) and 0.6 U/mL Dispase II (Roche, Bazel, Switzerland). Afterward, cells were pelleted by centrifugation at 300g for 5 min, incubated with 500–750 U of DNase (Roche) for 15–20 min, filtered through 70-μm filters, seeded onto a feeder (NIH/3T3-L1 fibroblasts, a kind gift from Dr. Vítězslav Bryja), irradiated (30 Gy), and cultivated on plastic dishes coated with 0.02% gelatin in phosphate buffered saline (PBS). Cells were cultivated in complete F medium containing Y-27632 according to the method of Liu et al. (13). All cells were cultivated in TPP (Trasadingen, Switzerland) or BD Falcon (BD Biosciences, San Diego, CA) cultivation dishes, flasks, and plates in a humidified incubator at 37°C in an atmosphere of 5% CO2. The cells were harvested after a brief incubation in 0.05% EDTA in PBS followed by trypsinization (0.25% w/v trypsin/0.53 mM EDTA in PBS), followed by counting using a Bürker chamber or CASY TT automatic cell counter (Roche). Cell suspensions were always filtered through sterile 100 μm syringe filters (Filcons, Steinfurt, Germany) before analysis.
Instrumentation, FACS, and Data Analysis
All measurements were performed on a FACSAria II Sorp 4L system (BD Biosciences) equipped with four lasers (excitation wavelengths: 355, 405, 488, and 640 nm, respectively). Details about the instrument configuration according to the MIFlowCyt guidelines (15) are shown in Supporting Information Table S1. Cell debris, doublets, and aggregates were excluded from analysis based on a dual-parameter dot plot in which the pulse ratio (signal area/signal high; y-axis) versus signal area (x-axis) was displayed. Dead cells were excluded from the analysis. The automatic compensation for multicolor immunofluorescence analysis was performed using BD CompBead beads (BD Biosciences) and particular antibodies. LIVE/DEAD Blue, Yellow, and Far Red dyes (Invitrogen) were compensated using the control sample prepared as a mix of dye-labeled dead cells and negative BD CompBead beads. Experiment measurements were performed using FACSDiva software (Version 6.1.3; BD Biosciences). Data analysis was performed using FlowJo software (Version 7.6.5, Tree Star). For sorting, a 100-μm nozzle (20 psi) was used, and the sorting rate was no more than 1000 events per second. List mode data are uploaded into FlowRepository database of flow cytometry experiments (http://flowrepository.org/id/FR-FCM-ZZ4Q)
Cell suspensions were sequentially diluted, and the cells were seeded by manual pipetting into a 96-well microplate at a dilution of 0.5 cells/well. Plates were incubated for 8 days under standard conditions.
For the ACCA, only viable cells were sorted using the automatic cell deposition unit according to the viability staining with propidium iodide solution (1 μg/mL, Sigma), LIVE/DEAD Blue, Yellow, or Far Red (1:500–1:1000). For the “single-cell” approach, cells were seeded at a dilution of 1 cell/well in replicates into 96-well [microscopic evaluation; 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), TPP plates] or 384-well plates (CyQUANT, Costar black plates with a black bottom; ATP, BD Falcon white plates with a clear bottom), and for the limiting dilution assay, cells were seeded in dilutions of 1–70 cells/well in replicates in 96-well plates.
Primary prostate tumor cells were sorted directly into 96-well TPP plates with the feeder cells as described previously (13). Both tubes with cells and plates with sown cells were incubated at 37°C during sample processing in the cell sorter. The sorted cells were cultivated for the next 8 (E cells) or 12 days (FHC and HCT-116 cells) under standard conditions; every third day, 50 (96-well plate) or 10 μL (384-well plate) of fresh complete cultivation medium were added to each well on the plate.
Microscopic Evaluation of Colonies
All microplate wells were examined under bright-field conditions with an Olympus IX-70 microscope and documented using an UPLanFl N 4×/0.13 objective, a Fluoview II CCD camera, and Analysis D software. A colony was defined to consist of at least 70 cells.
Cell Proliferation Assays
For MTT analysis in 96-well plates, a solution of MTT in PBS (2.5 mg/mL) was added to each well (10 μL). After incubation at 37°C for 2 h, the solution was replaced with 50 μL of 1% Triton X-100, and the plates were incubated at 37°C for 3 h. Absorbance was measured at 570/620 nm on an ASYS plate reader. The CyQUANT cell proliferation assay (Invitrogen) and ATP bioluminescence assay (Roche) were performed according the manufacturers' recommendations and analyzed using an Infinite M200 multimode reader (Tecan, Männedorf, Switzerland). The detection limits for assays were calculated as the mean value of the wells that contained one colony (>70 cells).
Trypsinized cE2 cells were incubated in 1% bovine serum albumin (BSA)/PBS buffer with APC-conjugated anti-CD133 (prominin-1; 1:11, Miltenyi Biotec, Bergisch Gladbach, Germany) or PE-conjugated anti-CD133 (1:100, Biolegend, San Diego, CA), and APC-Cy7-conjugated anti-CD44 (1:50, BioLegend) antibodies for 30 min at 4°C. Human HCT-116 cells were stained with PE-conjugated anti-CD133/1 (1:11, Miltenyi Biotec) or PE-conjugated streptavidin (1:2000, eBioScience) with anti-CD133/1 (AC133)-Biotin (1:11, Miltenyi Biotec) and APC-Cy7-conjugated anti-CD44 (1:50) antibodies. Human primary cancer cells derived from post-surgery biopsy specimens were stained with PerCP-eFluor 710-conjugated anti-CD49f (1:333, eBioscience, San Diego, CA), APC-Cy7-conjugated anti-CD44 (1:50), and unlabeled anti-Trop-2 (1:20, R&D Systems, Minneapolis, MN). For Trop-2 detection, Alexa Fluor 647 donkey anti-mouse secondary antibody (1:1000, Invitrogen) was used. Appropriate isotype antibodies were used for nonspecific binding analysis (rat IgG1 labeled with APC (1:11, BioLegend), rat IgG1 labeled with APC-Cy7 (1:50, BD Pharmingen), rat IgG2b kappa labeled with APC/Cy7 (1:50, BioLegend), mouse IgG1 PE (1:11, Miltenyi Biotec), rat IgG2a PerCP/eFluor 710 (1:333, eBioscience), or PE-conjugated streptavidin (1:2000, eBioScience)). Afterward, the samples were incubated with LIVE/DEAD Blue, Yellow, or Far Red Fixable Dead Cell Stain (1:500–1:1000) in PBS without BSA.
Real-Time Cell Analysis
Acea 96-well E-plates® and an xCELLigence real-time cell analysis (RTCA) SP system including RTCA Software v1.2 (both Roche) were used to monitor cell characteristics including cell number, adhesion, viability, and morphology, and obtain information about the biological status of the cells in real time by measuring the electrical impedance across microelectrodes integrated on the bottom of its special tissue culture plates (16, 17). First, a standard background measurement was performed using 200 μL complete culture medium. Immunolabeled cE2 and HCT-116 cells were seeded directly into the wells (in multiplicates of ≥4 wells for each subpopulation using a cell sorter (70 cells/well). The cell index (normalized electrical impedance) was monitored continually for a period of 312 h.
The clonogenic capacity for single-cell ACCA was calculated as the percentage of wells with colonies from the total number of sown cells. For limiting dilution analysis, the clonogenic capacity was estimated using extreme limiting dilution analysis (ELDA) software [(18), http://bioinf.wehi.edu.au/software/elda/]. The data were analyzed by STATISTICA for Windows (StatSoft, Prague, Czech Republic) using Wilcoxon's matched pairs test (comparison of results of clonogenic assays), t-test, one-way analysis of variance, or Mann-Whitney's U-test (if the data variances were nonhomogenous).
The Clonogenic Potential of Both Prostate and Colon Epithelial Cells Is Not Affected by Sowing Cells Using Electrostatic Droplet Sorting
In the present study, we introduced a straightforward approach for analyzing clonogenic capacity in vitro. In principle, we used the simple application of the capability of a droplet sorter equipped with an automatic cell deposition unit to precisely and aseptically seed single live cells in defined dilutions directly into microwell plates. First, we wanted to analyze whether the single-cell sorting of prostate and colon epithelial cells would affect their cloning capacity (plating efficiency) in vitro. All tested prostate and colon epithelial cell lines that were automatically sorted as single cells formed colonies with various morphologies (Fig. 1). We wanted to compare the method of cell seeding using a sorter with an ordinary seeding procedure (using an automatic micropipette) regarding its influence on clonogenic capacity in vitro. Therefore, we used identical suspensions of various trypsinized cell lines for either cell sorter seeding in the single-cell mode (1 cell/well, A-S) or manual pipetting (0.5 cells/well, M) into 96-well microplates in multiplicates. Each well on the microplate was analyzed using bright-field microscopy after 8 days of cultivation, and the clonogenic capacity was determined for each cell line and method. The results of at least five independent experiments indicated that the determined clonogenic capacity of each particular cell line (E2, E4, cE1, and cE2) was not significantly affected by the methods of cell sowing (Fig. 2A). Furthermore, we wanted to compare the single-cell mode of the automatic cell cloning assay with another approach using a similar principle, sowing cells at different dilutions (from 1 to 70 cells/well) using a cell sorter (A-LD) and estimate the clonogenic capacity of cell lines using an ELDA application (18). E2, E4, cE1, and cE2 cells were sown using a cell sorter at different dilutions (1, 2, 4, 6, 8, 12, 16, 20, 24, 35, 50, and 70 cells/well) in multiplicates into 96-well microplate. All wells in the microplates were examined under the microscope after 8 days of cultivation, and the counts of wells containing at least one colony were used to estimate the clonogenic capacity of cells using an ELDA application (http://bioinf.wehi.edu.au/software/elda/index.html). Using ELDA, the clonogenic capacity of E cell lines sown at different dilutions did not differ significantly from the results obtained using single-cell sorting (Fig. 2B).
Our results demonstrated that the use of a cell sorter to aseptically sow single live cells directly into microplate wells could be applied to the ACCA. When we compared the manual and automatic clonogenic assays, we found that instrumental cell sorting did not affect the clonogenic capacity of the tested cell lines. Moreover, we demonstrated that it is possible to modify the design of the ACCA and sort the cells either in the single-cell mode or according to the dilution and estimate clonogenic capacity using an ELDA application.
The CyQUANT and ATP assays, but not the MTT Assay, Display Sufficient Sensitivity for Endpoint Analysis of Clonogenic Capacity
Calculating the number of colonies in the microwell plate format using manual microscopic analysis in individual wells has relatively low throughput and is not applicable for 384-well formats. As high-throughput microscopic systems are not present in most laboratories, we wanted to test whether it is possible to use other cell-based assays for indirect endpoint determination of the clonogenic capacity of sorted cells in vitro. For comparison, we chose different assays to quantify cell proliferation and metabolic activity: MTT (colorimetric analysis of metabolic activity), CyQUANT (fluorescent quantification of DNA), and ATP bioluminescence assays. We wanted to reveal which method exhibits the highest sensitivity and signal-to-noise ratio to identify the best method for discriminating colony-negative and colony-positive wells. Different E cell lines were sorted by the number of cells per well (from 1 to 70 cells per well) in multiplicates into 96-well plates for MTT assays or 384-well plates for CyQUANT and ATP bioluminescence assays. After 8 days of cultivation, all tested assays were performed as described in the Materials and Methods section. The threshold for MTT and CyQUANT assays was defined as the median signal acquired from the wells with one colony (70 cells). Because of the high signal-to-noise ratio of the ATP bioluminescence assay, it was not necessary to set a threshold value for this particular method. A comparison of all three methods using four prostate epithelial cell lines demonstrated that the MTT assay has a low signal-to-noise ratio, and its sensitivity is not sufficient for the precise indirect analysis of clonogenic capacity. Conversely, both the CyQUANT and ATP bioluminescence assays exhibited high signal-to-noise ratios, great sensitivity, and dynamic ranges sufficient for the accurate indirect analysis of clonogenic capacity when they are combined with the automatic sowing of cells using a sorter (Fig. 3). Using both assays, in most of the tested cell lines, we were able to distinguish colony-positive and colony-negative wells when colonies arose from a single sorted cell after 8 days of cultivation.
The ACCA Revealed Increased the Clonogenic Capacity of Cancer Stem-Like Cells
The identification and functional analysis of rare tumor cell subpopulations, especially CSCs, is challenging. In this study, we used our experimental model of prostate and colon cancer cell lines to confirm that the ACCA is a flexible assay that is also applicable for characterizing phenotypically defined populations of cancer cells. Both cE2 and HCT-116 cells were immunolabeled with appropriate anti-CD44 and anti-CD133 (anti-prominin-1) fluorescent antibodies, and live subpopulations of CD44+/CD133low (low: bottom 25% of CD133-positive cells) or CD44+/CD133high (high: top 25% of CD133-positive cells) cells were sorted in the single-cell or dilution rank mode into 96-well microplates. The number of colony-positive wells was determined using bright-field microscopy, and the clonogenic capacity for both subpopulations of cE2 and HCT-116 cells was calculated (single-cell mode, A-S) or estimated using the ELDA application (dilution rank mode, A-LD) (Fig. 4). Our data revealed the significantly increased clonogenic capacity of CD44+/CD133high cells compared to that of CD44+/CD133low cells in both cE2 and HCT-116 cells. Next, we wanted to determine whether the CyQUANT and ATP bioluminescence assays, which were previously identified as sufficient readouts for indirect determination of clonogenic capacity, would confirm this observation. Subpopulations of CD44+/CD133high and CD44+/CD133low cE2 cells were sorted in the single-cell mode as described previously. The CyQUANT and ATP bioluminescence assays were performed after 8 days of cultivation. Both assays revealed the increased clonogenic capacity of CD44+/CD133high cells compared to that of CD44+/CD133low cells in the cE2 cell line (Figs. 5A and 5B) and HCT-116 (Figs. 5C and 5D).
Label-free methods are exhibiting increasing potential in various areas of cellular biology including cancer stem cell research. Therefore, we wanted to combine the automatic sowing of phenotypically defined subpopulations of cancer stem-like cells using a sorter with a real-time cell analyzer (xCELLigence). The xCELLigence system monitors cellular events in real time without the incorporation of labels. It measures the electrical impedance across interdigitated microelectrodes integrated on the bottom of a tissue culture microplate (E-plate). The impedance measurement provides quantitative information about the biological status of cells, including cell number, viability, and morphology. Subpopulations of CD44+/CD133high and CD44+/CD133low cE2 respectively CD133low and CD133high HCT-116 cells were sorted (70 cells/well) directly into the E-plate. Then, the cell index was monitored in real time for 312 h (Fig. 5E). Analysis of the obtained data revealed that CD133high cells reached the half-maximal cell index (CI50) value in significantly less time than CD133low cells (213 vs. 272 h in cE2 cells; 265 vs. >312 h in HCT-116 cells). We can conclude that this long-term experiment revealed that the increased clonogenic capacity of CD133high cells is also associated with the increased proliferation capability of this cell subpopulation in both tested cell lines. Taken together, we proved that ACCA is a flexible assay that is applicable for characterizing phenotypically defined cell subpopulations including cancer stem-like cells. We demonstrated that CD44+/CD133high cancer cells exhibit increased clonogenic capacity. Moreover, for the first time, we have demonstrated that it is possible to combine direct sorting (sowing) of a limited number of immunolabeled cells with long-term real-time label-free analysis of cell proliferation.
Application of the ACCA to a Primary Prostate Cancer Sample
Next, we wanted to confirm that it is possible to use the ACCA in both in vitro cultivated stable cancer cell lines and primary cancer samples. As a proof of principle, we used samples of prostate cancer tissue obtained from patients with advanced prostate cancer with perineural invasion. The cells were isolated and cultivated as described in the Materials and Methods section. Cells were labeled with fluorescent antibodies against TROP-2, CD49f, and CD44, and TROP-2+/CD49f+/CD44low and TROP-2+/CD49f+/CD44high cells were seeded using a cell sorter in the single-cell mode (1 cell/well) into 96-well microplates with irradiated 3T3 feeder cells. The number of colony-positive wells was determined using bright-field microscopy. Our analysis did not reveal any significant difference in clonogenic capacity between TROP-2+/CD49f+/CD44low and TROP-2+/CD49f+/CD44high cells (data not shown); however, we clearly demonstrated that the ACCA is a simple and straightforward approach that is applicable to primary cancer samples (Fig. 6).
The clonogenic assay is a basic tool that was originally introduced in 1956 by Puck and Markus (11) to assess the effects of radiation on the clone-forming ability of single cells plated in vitro on feeder cells (12). However, the principle of the analysis that determines the ability of a cell to proliferate indefinitely and generate a large colony or clone has been widely applied in various areas of biological research including cancer research.
The principle of the method combining the clonogenic assay and flow cytometric cell sorting was described for the first time by Park et al. in 1987 (19). The authors declared that it was possible to determine the clonogenic capacity of aseptically sorted live Chinese Hamster Ovary and human leukemic bone barrow cells based on the DNA distribution. However, this application remained overlooked even as flow cytometry and sorting instrumentation significantly improved in the last decade. FACS represents a sensitive, specific, and robust approach for the phenotypic analysis and purification of a variety of cell populations including CSCs (9). In this study, we established the ACCA in a microwell format based on the principle of using an automatic cell deposition unit and a FACS system. This approach can inherently utilize the advantages of sorting strategies based on the multiparametric detection of cell surface markers in live cells. Our results indicated that the novel ACCA is a straightforward approach for determining the clonogenic capacity of different subpopulations of cancer stem-like cells identified in both cell lines and patient samples.
First, we examined whether it is possible to use an automatic cell deposition unit to aseptically sow various mouse and human cancer cell lines directly into microplates. Our results revealed that in our hands, the application of a 100-μm nozzle in the cell sorter does not significantly affect the post-sorting viability of cells (data not shown) or their capacity to form anchorage-dependent colonies in vitro. When we compared manual and automatic approaches for clonogenic assays, we determined that instrumental cell sorting does not affect the clonogenic capacity of the tested cell lines. Moreover, we demonstrated that it is possible to modify the design of the ACCA and sort cells either in a single-cell or dilution rank mode and estimate the clonogenic capacity of cells using an ELDA application (18).
In principle, the best approach for analyzing and quantifying colony-forming capacity in the microplate format is using HTS microscopic readers. Although HTS microscopy is an important research tool, it is not always available. Therefore, we tested and compared simple biochemical methods for the endpoint analysis of colonies in microplates based on the principle of cell quantification (MTT, CyQUANT, ATP). A comparison of these methods using four prostate epithelial cell lines demonstrated that the MTT assay has a low signal-to-noise ratio, and its sensitivity is not sufficient for the precise indirect analysis of clonogenic capacity. By contrast, both the CyQUANT and ATP bioluminescence assays had high signal-to-noise ratios, great sensitivity, and dynamic ranges sufficient for the accurate indirect analysis of clonogenic capacity when they are combined with automatic cell sowing using a sorter. Using both of these assays, we were able to distinguish colony-positive and colony-negative wells for most of the cell lines tested when colonies arose from a single sorted cell in the 384-well format. These result indicated the possibility of simply upscaling this approach for high-throughput screening using such basic biochemical readouts.
Next, we wanted to prove that the ACCA is a flexible assay that is applicable for characterizing phenotypically defined cell subpopulations including cancer stem-like cells. As an example, we chose two cell lines, mouse prostate cancer cE2 cells and human colon cancer HCT-116 cells, which both exhibit positivity for two typical markers of prostate and colon cancer stem cells: CD44 and CD133 (20, 21). The ACCA demonstrated that the CD133high subpopulations in both cell lines display increased clonogenic capacity. These observations are in agreement with a previously published study by Jacksch et al. demonstrating the higher colony-forming capacity of CD133high Caco2 colon cancer cells (22).
Label-free RTCA based on the detection of cellular impedance had great potential in both basic and applied research (16, 17, 23, 24). In this study, we revealed for the first time that it is possible to combine the direct deposition of a limited number of immunolabeled cells with long-term, real-time, label-free analysis of cell proliferation. Interestingly, our data obtained using the RTCA system demonstrating the increased proliferative capability of CD44+/CD133high and CD133high cancer cells are in agreement with the increased clonogenic capacity of CD44+/CD133high subpopulations of cE2 and CD133high in HCT-116 cells respectively.
Furthermore, we applied this approach to primary patient samples. Primary prostate cancer cells did not display CD133 expression; therefore, we decided to compare the clonogenic capacity of the cells based on the expression of other described cancer stem cells markers, TROP-2+, CD49f, and CD44, using the ACCA. Although our analysis did not reveal any significant difference in clonogenic capacity between TROP-2+/CD49f+/CD44low and TROP-2+/CD49f+/CD44high cells, we were able to clearly prove that the ACCA is a simple and straightforward approach that is applicable to primary cancer samples. Moreover, we have confirmed that ACCA is in principle an applicable approach for isolating and expanding single cell-derived clones of phenotypically defined primary cells when it is combined with a highly effective protocol for establishing cell lines from primary samples as described by Liu et al. (13).
In summary, we have determined that the direct sowing of cells using a cell sorter equipped with an automatic deposition unit into microwell plates is a useful approach for determining the anchorage-dependent clonogenic capacity of cells or obtaining clones from heterogeneous patient samples via the expansion of single phenotypically defined cells. The major advantages of the described ACCA are its simplicity, rapidity, and straightforwardness, which could reduce the steps required when cells are manipulated. This assay could be simply scaled-up and fully automated for use in high-throughput drug screening in combination with HTS microscopy.
The authors would like to thank Prof. Pradip Roy-Burman for his kind gift of E2, E4, cE1, and cE2 cells; Prof. Bert Vogelstein for his kind gift of HCT-116 Neo cells; Dr. Vítězslav Bryja for his kind gift of NIH 3T3-L1 cells; Petra Ovesná for help with statistical analysis; and Miluše Andrysíková, Martina Urbánková, Monika Levková, Jaromíra Netíková, and Kateřina Svobodová for their superb technical assistance.