Understanding the role of cancer stem-like cells in tumor recurrence and resistance to therapies has become a topic of great interest in the last decade. This article describes the isolation and characterization of the sub-population of cancer stem-like cells from head and neck squamous carcinoma cell lines (HNSCC).
Despite advances in the understanding of head and neck squamous cell carcinomas (HNSCC) progression, the five-year survival rate remains low due to local recurrence and distant metastasis. One hypothesis to explain this recurrence is the presence of cancer stem-like cells (CSCs) that present inherent chemo- and radio-resistance. In order to develop new therapeutic strategies, it is necessary to have experimental models that validate the effectiveness of targeted treatments and therefore to have reliable methods for the identification and isolation of CSCs. To this end, we present a protocol for the isolation of CSCs from human HNSCC cell lines that relies on the combination of two successive cell sortings performed by fluorescence activated cell sorting (FACS). The first one is based on the property of CSCs to overexpress ATP-Binding Cassette (ABC) transporter proteins and thus exclude, among others, vital DNA dyes such as Hoechst 33342. The cells sorted with this method are identified as a "side population" (SP). As the SP cells represent a low percentage (<5%) of parental cells, a growing phase is necessary in order to increase their number before the second cell sorting. The next step allows for the selection of cells that possess two other HNSCC stem cell characteristics i.e. high expression level of the cell surface marker CD44 (CD44high) and the over-expression of aldehyde dehydrogenase (ALDHhigh). Since the use of a single marker has numerous limitations and pitfalls for the isolation of CSCs, the combination of SP, CD44 and ALDH markers will provide a useful tool to isolate CSCs for further analytical and functional assays requiring viable cells. The stem-like characteristics of CSCs was finally validated in vitro by the formation of tumorispheres and the expression of β-catenin.
Head and neck squamous cell carcinoma (HNSCC) is a common malignancy worldwide and despite progress in current treatments, patients with advanced disease have a poor prognosis. The overall 5 year survival rate of the patient is around 30% despite the combination of therapeutic approaches including surgery, chemo-radiotherapy and targeted-therapies. Recent studies attribute local recurrence and distant metastasis to the survival of cancer stem-like cells (CSCs) following anticancer therapies1. There is accumulating evidence supporting the existence of cells presenting stem cells properties (undifferentiated status, self-renewal and differentiation capacities, and telomerase activity) in various solid tumors including breast, brain, prostate, lung, colon, pancreas, liver and skin2-10. However, the origin of CSCs remains unclear11,12. They may result from the malignant transformation of normal stem cells3,13 or dedifferentiation of tumor cells that acquire CSCs-like features14,15. Therefore, understanding distinctive pathways relating to CSCs will provide insight into early diagnosis and treatment of resistant HNSCC.
It has been proposed that CSCs also possess resistant phenotypes that evade standard chemotherapy and radiotherapy, resulting in tumor relapse compared to the bulk of tumor cells16-19 and are localized into hypoxic niches20. Numerous factors have been proposed to explain these resistances of CSCs, such as propensity to quiescence, enhanced DNA repair, up-regulated cell cycle control mechanisms, and free-radical scavenging21. Moreover, several oncogenic molecular pathways may be specifically activated in CSCs17. In order to improve knowledge of CSCs for further targeted-therapies, we need reliable methods for the identification and isolation of CSCs, owing to the heterogeneity of stem cell-related markers in various types of cancers22.
In HNSCC, stem-like tumor-initiating cells have been isolated from primary patient tumors by sorting cells expressing different CSC biomarkers (such as drug efflux transporters expression23, CD44high, CD24low CD133high, c-Met+ phenotype10,24,25, or ALDHhigh activity26) or cultivating primary patient tumor to form squamospheres that have CSC properties. Nevertheless, the number of squamospheres decreases dramatically after two passages, thus giving a small sample size for further characterization studies27. Therefore, in vitro assays starting from well-established cell lines is an easier solution to design experiments in order to improve knowledge of CSCs.
The aim of this article is to propose a method to isolate CSCs from HNSCC cell lines using multiparametric flow cytometric analysis and cell sorting. The expression of CD44 in correlation with several CSCs properties including ALDH activity, Side Population (SP) phenotype, spheroid formation ability and tumorigenicity are used to isolate and characterize this sub-population of CSCs. CD44, a cell-surface glycoprotein, is involved in cell adhesion and migration. CD44 is highly expressed in many solid tumors CSCs28, including in head and neck cancer models29-31. Moreover, CD44high cells can generate in vivo a heterogeneous tumor whereas CD44low cells cannot10. The SP assay is based on the differential potential of cells to efflux the Hoechst dye22 via the ATP-binding cassette (ABC) family of transporter proteins overexpressed within the CSC membrane. This assay includes the use of ABC transporter inhibitors such as verapamil in control samples. Aldehyde dehydrogenase (ALDH) is an intracellular enzyme that is involved in converting retinol to retinoic acid during early stem cell differentiation25,26. Cells that exhibit high ALDH activity show stem-like cell behavior in HNSCC26 and a very few number of ALDHhigh cells are able to generate tumors in vivo26,32.
The combination of these markers and properties was successfully used by Bertrand et al. to study the resistance in vitro and in vivo of these CSCs to photon and carbon ion radiation19. Their results clearly showed that the combination of various cell markers and properties are more selective for useful studies on HNSCC CSCs populations than single-marker approaches.
All animal procedures were performed according to local guidelines on animal care. All the details of this study were approved by the CECCAPP, a French ethics committee.
1. Selection of a Side Population (SP) by the Hoechst Dye Efflux Assay
2. Selection of the CD44high/ALDHhigh Sub-population in the Side Population Sorted
3. Cell Culture Method
4. Confirmation of Tumor Potential and CSC Characteristics
The isolation of CSCs from HNSCC cell lines required two successive sorting because of the very low percentage of CSCs in the parental cell line. The first sorting was based on the ability of CSCs to exclude the Hoechst dye due to drug efflux transporters. This resulted in acquisition of 1-5% of the total cell population sorted (Figure 1). During the Hoechst dye negative cell sorting, check the size and granulation of sorted cells by looking at the FSC-A versus SSC-A dot plot (Figure 1A). Then, discriminate doublets and cells fragments by using the SSC-W versus SSC-H dot plot and selecting the population P1 (Figure 1B). On this P1 population, create a Hoechst Red-A versus Hoechst Blue-A dot plot. With the tube labeled "Hoechst", the SP appears as a side arm on the left from the main cells population (Figure 1C). This population must disappear when they are treated with Verapamil (Figure 1D), an inhibitor of ABC transporters. The PI staining allows the exclusion of PI-positive dead cells because this population is under-scale on the Hoechst Red-A scale (Figure 1C and 1D, blue ellipses).
The second cell sorting was based on the high expression of the CD44 receptor and high ALDH enzyme activity which allowed the acquisition of 0.5-2% from the SP cells previously sorted (Figure 2). Before the sorting, various controls were used. The first ones are the "ALDH" and the "ALDH + DEAB" tubes needed in order to place the first gate on FITChigh cells (Figures 2A and 2B): ALDHhigh cells were gated on the FITC-A versus APC-A dot plot using the "ALDH" tube (Figure 2A). The good position of the gate was checked using the "ALDH + DEAB" tube: as DEAB inhibits ALDH, positive cells must disappear from the gate (Figure 2B). If they don't, change the reaction conditions (by increasing the amount of DEAB for example). The second control is the "CD44-APC" and the "IgG1-APC" tubes which allowed to position the second gate on APChigh cells (Figures 2C and 2D) using the cells stained with the CD44-APC antibody (Figure 2C). This population must disappear with the control tube which contained IgG1-APC cells (Figure 2D). If it does not, the bond with the antibody is not specific and BSA 0.5% should be added into buffer 1 from the ALDH detection kit during the antibody reaction. Finally, the third control concerns the "ALDH and CD44-APC" tube and "ALDH, DEAB and CD44-APC" tubes (Figures 2E, 2F, 2G and 2H). The double staining ALDH/CD44 tube is used to position the last gate on the double positive cells (Figure 2E) and the same tube treated with DEAB is a control to verify that ALDHhigh cells disappear (Figure 2F).
This protocol is used to sort CSCs from the SQ20B and the FaDu cell lines. When the sorting is done for the first time on a new cell line, in order to ensure that sorted cells have stem-like cells properties, it is necessary to confirm their tumor potential. One of the stem-like cell property is its ability to form tumorspheres in vitro in a FSC free medium. Under this condition, only cancer stem-like cells can survive and proliferate (Figure 3A). Moreover, qPCR experiments show a high expression of β-catenin (marker of stem-like characteristic) in CD44high/ALDHhigh cells (Figure 3B), as well as Bmi-1 and Notch19. Finally, CD44high/ALDHhigh cells are also able to form tumors when injected in low quantities as compared with CD44low/ALDHlow cells19.
Figure 1: Isolation of a side population excluding the Hoechst dye. (A) FSC-A versus SSC-A dot plot. (B) SSC-W versus SSC-H dot plot. (C, D) Hoechst Red-A versus Hoechst Blue-A dot plot using the "Hoechst" tube (C) or with the "Hoechst and Verapamil" tube (D). Please click here to view a larger version of this figure.
Figure 2: Sorting of CD44high/ALDHhigh cells from Hoechst dye negative cells. FITC-A versus APC-A dot plot using the "ALDH" tube (A), the "ALDH + DEAB" tube (B), the "CD44-APC" tube (C), the "IgG1-APC" tube (D), the "ALDH and CD44-APC" tube (E and G) or the "ALDH, DEAB and CD44-APC" tube (F and H). Figures A, B, C, D, E and F were obtained using SQ20B cell line. Figures G and H obtained using FaDu cell line. Green indicates CD44low/ALDHlow cells (Q3); dark blue indicates CD44low/ALDHhigh cells (Q1); purple indicates CD44high/ALDHlow cells (Q4); light blue indicates CD44high/ALDHhigh cells (Q2). Please click here to view a larger version of this figure.
Figure 3: Confirmation of tumor potential of CD44high/ALDHhigh cells. Sorted CD44high/ALDHhigh cells were able to form tumorspheres in vitro (A) in a poor-FCS media. Scale bar, 25 µm. Furthermore, CD44high/ALDHhigh overexpress β-catenin, a marker of tumorigenicity (B). This quantification has been performed by qPCR and error bars represent SD. Please click here to view a larger version of this figure.
Number of cells sorted | Culture flask type | Trypsin Volume (ml) | Culture Medium Volume (ml) |
10,000-200,000 | 1 well of a 6-well plate or a 3.5 cm petri dish | 0.5 | 2 |
200,000-1,000,000 | 1 T25 culture flask | 1 | 4 |
>1,000,000 | 1 T75 culture flask | 2 | 10 |
Table 1: Culture flask type to use according to the number of cells sorted. Details of the culture flask size for approximate number of cells sorted are given. The volume of trypsin and volume of medium required for the culture flask type are also provided.
This protocol describes a reliable method for the successful isolation of CSCs from a specific cell line that is applicable to other HNSCC cell lines. Isolated head and neck CSCs are then suitable for further molecular characterization in vitro and functional validation by transplantation in immunodeficient mice19. However, some modifications can be tested depending on the side population or the CD44high/ALDHhigh percentages present in the parental cell line. For example, if the percentage of cells in the side population is too low in a particular cell line, the CD44high/ALDHhigh sorting can be performed directly. The markers used in this study may be replaced by other markers appropriate to the cell line studied such as CD133high 34 or CD10high 35.
This protocol is based on two cell sortings. Three color sorting with SP + CD44 + ALDH was not possible with the cell lines tested. First, the parental cell lines tested here present less than 5% of SP and less than 10% CD44high/ALDHhigh cells in the SP. Therefore, SP/CD44high/ALDHhigh represent less than 0.5% of the parental cell line. Hence, a first SP sorting is necessary in order to enrich the population in CD44high and/or ALDHhigh cells. Second, to sort 1% of the parental cell line, it takes 5 hr to obtain approximately 300,000 cells. Therefore, if a 3 color cell sorting is undertaken, the quantity of cells collected will be very low. Furthermore, longer sorting is not recommended as it may affect cell viability.
Owing to the selectivity of this isolation protocol, the main limitation is the small number of CSCs obtained. This could be problematic to perform further experiments, since it is not recommended to use them after 3 passages because of the rapid loss of CD44 and ALDH markers. Furthermore, before each new experiment, the percentage of CD44high/ALDHhigh cells still present in the cell suspension should be tested in order to check the number of CSCs that has differentiated.
It is imperative to prepare Verapamil hydrochloride solution and culture medium containing EGF just before use as these molecules are very unstable. A stock solution of EGF at 20 mg/L is stable for three months at -20 °C. Variations of the Hoechst protocol exist, but the final dye concentration commonly used is 5 µg/ml. Moreover, during the first sorting, different culture media compositions should be tested according to the cell line used. Once the sorting conditions are validated, it is necessary to check the properties of the sorted cell population using the different methods (section 4).
Cell surface markers, ALDH activity and ability to efflux vital dyes have been already used individually in the literature to isolate CSCs from HNSCC. However, the protocol described here has the unique advantage of using combinations of various markers in order to achieve high specificity in CSCs isolation from HNSCC cell lines. Moreover, the CD44 sorting could be realized with an antibody anti-CD44 conjugated with magnetic micro-beads and sorted with a magnetic column36, but this method is only applicable to cell surface markers and cannot be used for an ALDH sorting, preventing double sorting. Another method used to obtain CSCs is the tumorsphere culture from primary tumors37 or xenografts38. However, the acquisition of these primary tumors or xenografts are associated with ethical constraints.
Since this method isolates viable HNSCC CSCs, they can be used (after checking their tumorigenicity) in a number of experiments that measure the physiological function of these cells. It therefore allows assessment of the behavior of CSCs following different therapeutic approaches (radiotherapy, chemotherapy). It also allows the study of various biological parameters such as migration/ invasion, DNA repair, cell signaling, etc. Hence, obtaining CSCs from different cell lines is an attractive choice for investigating CSCs properties.
The authors have nothing to disclose.
We thank Thibault Andrieu and Sebastien Dussurgey from the Flow Cytometry Platform of UFR BioSciences Gerland-Lyon-Sud (UMS3444/US8) for their advice and help during our sorting. This work was achieved within the scientific framework of ETOILE and Labex-PRIMES (ANR-11LABX-0063).
Fetal Calf Serum Gold | GE Healthcare | A15-151 | |
Hydrocortisone water soluble | Sigma-Aldrich | H0396-100MG | |
Penicillin/Streptomycin 100 X | Dominique Dutscher | L0022-100 | |
DMEM | Gibco | 61965-026 | |
F12 Nut Mix (1X) + GlutaMAX-I | Gibco | 31765-027 | |
EGF | Promega | G5021 | The solution must be prepared just before use because it is very unstable |
Heparin | StemcellTM Technologies | 7980 | |
B-27 Supplement (50X), minus vitamin A | Gibco | 12587-010 | |
Hoechst 33342 | Sigma-Aldrich | 14533 | Corrosive, acute toxicity (oral, dermal, inhalation) category 4 |
Verapamil hydrochloride | Sigma-Aldrich | V-4629 | Acute toxicity (oral, dermal, inhalation) category 3 |
Propidium Iodide | Sigma-Aldrich | P4170 | Acute toxicity (oral, dermal, inhalation) category 4 |
ALDEFLUOR Kit | Stem Cell | 1700 | |
CD44-APC, human antibody | Miltenyi Biotech | 130-095-177 | |
IgG1-APC, human antibody | Miltenyi Biotech | 130-093-189 | |
Z1 coulter particle | Beckman Coulter | 6605698 | |
Optical microscope | Olympus | CKX31 | |
SQ20B cell line | Gift from the John Little’s Laboratory | – | |
FaDu cell line | ATCC | HTB-43 | |
Low anchorage plates | Thermo Fischer Scientific | 145383 | |
BD FACSDiva software v8.0.1 | BD Biosciences | – |