Background
The biological and clinical significance of intra-tumoral heterogeneity have been increasingly recognized [
1]. Some degree of intra-tumoral heterogenicity is believed to be resulted from clonal evolution, a process in which individual cancer cell clones gradually acquire an increasing number of genetic defects and the subsequent selection of cell clones carrying highly aggressive or metastatic phenotype [
2]. Intra-tumoral heterogeneity is also likely linked to the concept of cancer stem cells (CSCs), which represent a very small population within the tumors that are capable of self-renewal as well as generation of various elements of the tumors [
1,
2]. In recent years, accumulating evidence has suggested that some of the bulk tumor cells can acquire cancer stemness under certain circumstances or specific stimulations [
1‐
3]. This phenomenon is encompassed under the concept of “acquired stemness’ or cancer cell plasticity [
4]. Thus far, cancer cell plasticity has been largely described and studied in epithelial and neurogenic malignancies. For instance, MCF7, an estrogen receptor-positive breast cancer cell line, was found to acquire higher levels of tumorigenic potential and chemoresistance upon oxidative stress [
5]. In another study, it was found that hypoxic stress can result in the enrichment of CSCs in gliomas, which can be identified by means of their high CD133 expression [
6]. To our knowledge, cancer cell plasticity has not been described in hematologic malignancies.
ALK-positive anaplastic large-cell lymphoma (ALK+ALCL) is a distinct form of T-cell non-Hodgkin lymphoma recognized in the World Health Organization Classification of Hematopoietic Neoplasms [
7]. The majority of these tumors are found to carry the characteristic reciprocal chromosomal translocation involving
ALK and
NPM, and the resulted fusion gene product,
NPM-ALK [
7]. NPM-ALK has been shown to be the key oncogenic driver of ALK+ALCL [
8]. A large body of experimental data has suggested that NPM-ALK mediates its oncogenic potential by exerting its constitutively active tyrosine kinase activity on a host of cellular signaling proteins, including those in the JAK/STAT, MAPK/ERK, and PI3K/AKT pathways, resulting in their inappropriate activation [
8]. In our prior studies of ALK+ALCL, we had identified two distinct cell subsets in ALK+ALCL based on their differential response to a commercially available Sox2 reporter, which key components consist of the Sox2 regulatory region 2 (SRR2) as well as two gene reporters (
Green Fluorescence Protein, GFP and
Firefly luciferase) [
9]. Specifically, we found that reporter responsive (RR) cells are significantly more tumorigenic and stem-like than reporter unresponsive (RU) cells. In a recently published manuscript, we also have reported that the Wnt/β-catenin/MYC/Sox2 signaling axis is the key regulator of the RU/RR dichotomy in ALK+ALCL, with this axis being substantially more active in RR cells than RU cells [
10]. Due to the inclusion of the
GFP gene in the Sox2 reporter, RU and RR cells present in ALK+ALCL cell lines can be easily detected and purified by using flow cytometry.
In this study, we found that a fraction of RU cells can convert to RR cells upon oxidative stress, and this conversion was associated with the acquisition of cancer stem-like features. We believe that these observations exemplify cancer cell plasticity in a hematologic malignancy. Our study results also have provided further support that the Wnt/β-catenin/MYC/Sox2 axis is the key regulator of cancer stemness in ALK+ALCL.
Methods
Cell lines and chemicals
Karpas 299, an ALK+ALCL cell lines, was a kind gift from Dr. Marshall Kadin (Boston, MA). SupM2, another ALK+ALCL cell lines, was purchased from Deutsche Sammlung von Mikrooorganismen und Zellkulturen GmbH (DSMZ), Germany (catalog number, ACC 509). Karpas 299 and SupM2 cell lines were grown in RPMI 1640 medium (Invitrogen, Life Technologies, Grand Island, NY) with supplementations of 10% fetal bovine serum (Invitrogen), 1% Penicillin/Streptomycin (ThermoFisher Scientific, Burlington, Ontario, Canada) and 2 μg/mL puromycin dihydrochloride (Sigma Aldrich, St Louis, MO) in 5% CO2 atmosphere at 37 °C. Chemicals used in this study, including hydrogen peroxide (H2O2), doxorubicin, N-acetyl-L-cysteine (NAC), iodonitrotetrazolium chloride, 10074-G5 and quercetin, were purchased from Sigma Aldrich.
Flow cytometric isolation of RU and RR cells
RU and RR cells derived from SupM2 and Karpas 299 were purified with the use of a flow cytometry cell sorter (Becton Dickinson Biosciences, Franklin Lakes, NJ), and details of the method have been described previously [
10].
Luciferase reporter assay and flow cytometry
The luciferase reporter activity was measured by using the luciferase assay kit (Promega, Madison, WI), and the procedure was in accordance with the manufacturer’s protocol. The performance of flow cytometry has been detailed in a previous publication [
9]. A software (FCS Express 5) from De Novo Software (Glendale, CA) was used to analyze the generated flow cytometry data.
Trypan blue exclusion and MTS
Trypan blue exclusion and the MTS were used to quantify the number of viable cells (i.e. cell growth), and details of these assays have been previously described [
11].
Oxidative stress treatment
SupM2 or Karpas 299 cells growing in the log phase were seeded into 25 mL Corning™ cell culture flask (ThermoFisher Scientific Canada). For the H2O2 re-challenge experiments, cells were initially treated with two different doses of H2O2 (0.3 or 0.5 mM) for 3 days. On day 3, cells were subjected to centrifugation (50 X g, 10 min, room temperature), washed and cultured in fresh culture medium supplemented with either 0.3 mM or 0.5 mM H2O2 for two additional days (i.e. day 4 and day 5). At the end of the experiments, we performed luciferase reporter assay and flow cytometry to evaluate the luciferase activity and GFP expression, respectively. For the N-acetyl-L-cysteine (NAC) experiments, RU cells derived from SupM2 were treated with 0.3 mM H2O2 in the presence of 0–20 mM of NAC for 2 days, and these cells were assessed for their SRR2 reporter activity (i.e. luciferase activity and GFP expression). For the 10074-G5 or quercetin experiments, ALK+ALCL cells were treated with 5 μM of 10074-G5 or 50 μM of quercetin from day 4 to day 5 (24 h in total).
Short interfering RNA and transfection
Short interfering RNA (siRNA, SMART pool, Dharmacon, Lafayette, CO) against Sox2 was used to down-regulate Sox2. Cells transfected with scrambled siRNA (Dharmacon) were included as the negative controls. BTX Electro Square ECM830 (225 V, 8.5 milliseconds, 3 pulses) was used to transfect siRNA species into ALK+ALCL cells. The efficiency of siRNA knockdown of Sox2 was assessed by using Western blotting.
The method has been described previously [
10]. Primer sequences used in this study were listed in Table
1.
Table 1
The primer sets for qRT-PCR performed in this study
BCL9 | 5’-GGCCATACCCCTAAAGCACTC-3’ | 5’-CGGAAATACTTCGCTCCTTTT-3’ |
CTNNB1 | 5′- AAAGCGGCTGTTAGTCACTGG-3’ | 5′- CGAGTCATTGCATACTGTCCAT-3’ |
SOX2 | 5′- GCCGAGTGGAAACTTTTGTCG-3’ | 5′- GGCAGCGTGTACTTATCCTTCT-3’ |
WNT2B | 5′- GATCAAGATGGTGCCAACTTC-3’ | 5′- CCAAGACACAGTAATCTGGAGAG-3’ |
GAPDH | 5′- GGAGCGAGATCCCTCCAAAAT-3’ | 5′- GGCTGTTGTCATACTTCTCATGG-3’ |
Western blotting
Details of the method have been described previously [
11]. Antibodies reactive to phosphorylated ALK
Y1604 (catalog #3341), ALK (D5F3®), phosphorylated STAT3
Y705 (D3A7), STAT3 (124H6), phosphorylated MYC
S62 (E1J4K), MYC (D84C12), Sox2 (D6D9), and β-catenin (D10A8) were purchased from Cell Signaling Technology (Danvers, MA). Antibody reactive to β-actin (sc-130300) was purchased from Santa Cruz (Dallas, TX), and antibody against active β-catenin (8E7) was from Merck Millipore (Toronto, ON, Canada). Anti-mouse IgG (catalog #7076) and anti-rabbit IgG (catalog #7074) were from Cell Signaling. Densitometry data was analyzed using Imag J software (National Institutes of Health, Bethesda, MD), and the values were normalized to those of β-actin or γ-tubulin.
SRR2 probe binding assay
Details of this method have been described previously [
10]. The sequence of the SRR2 probe is as follows:
5‘-AAGAATTTCCCGGGCTCGGGCAGCCATTGTGATGCATATAGGATTATTCACGTGGTAATG-3′. The underlined sequence is the SRR2 consensus sequence.
The methylcellulose-based media were purchased from the R&D Systems Inc., (Minneapolis, MN). For each experimental group, we seeded 500 cells in each well, as described previously [
11]. After ~ 10 days of culture, we employed iodonitrotetrazolium chloride to stain the colonies. Only colonies with > 40 cells were counted. Triplicate experiments were done. Images of the colonies were acquired using the Alphalmager HP (ThermoFisher Scientific Canada).
Limiting dilution assay
ALK+ALCL cells, with or without H2O2 treatment, were seeded in RPMI1640 (Invitrogen) supplemented with 20% fetal bovine serum (Invitrogen), 1% penicillin/streptomycin and 2 μg/mL puromycin dihydrochloride (Sigma Aldrich) in 96-well low adherent plate (Corning). Ten different cell numbers, ranging from 1 to 1000 cells were used. After 7 days of culture, cells were stained with iodonitrotetrazolium chloride (Sigma Aldrich) for 24 to 48 h and images were acquired by using Alphalmager HP (ThermoFisher Scientific Canada). Cell spheres visible to naked eyes were counted. Triplicate experiments were used.
Statistical analysis
The generated data in all quantitative assays of this study was presented as mean ± standard deviation. GraphPad Prism 5 software (La Jolla, CA) was employed for statistical analysis. The statistical significance of the differences between two experimental groups of test samples was assessed by using Student’s t-test. Statistical significance is denoted by * (p < 0.05) or ** (p < 0.01).
Discussion
In several recent reviews, the concepts and significance of cancer cell plasticity and acquired cancer stemness have been eloquently explained [
2,
17,
18]. One of the key elements of these concepts is that the CSC is a dynamic phenotype, which can be acquired by non-stem cells if appropriate stimuli and microenvironment are provided. Virtually all of the published studies supporting these concepts are based on experimental data derived from malignant epithelial and neural-derived cells, and the induction of cancer stem-like features often involves the use of various chemotherapeutic agents, radiation therapy, hypoxia and oxidative stress [
2,
17,
19,
20]. For instance, the size of the CSC population in head and neck squamous carcinoma cells, detectable by their ALDH
high/CD44
high phenotype, was found to be significantly increased when these cells were exposed to cisplatin [
21]. In two other studies, human pancreatic cancer cells subjected to a fibrogenic microenvironment were found to acquire CSC-like features [
22,
23]. Hypoxia-inducible factors (HIFs), which exists in specific tumor microenvironment niche, is also known to promote metastasis, tumorigenicity and CSC-like features in certain human cancer models [
24,
25].
The biochemical basis of how cancer cells acquire stemness is not completely understood, although there is evidence that activation of specific signaling pathways or transcription factors is likely important. In this regard, Saijo et al. have reported that oxidative stress-induced de-differentiation of lung cancer cells into CSCs is mediated via the activation of HOXA5 and upregulation of Sox2 [
26]. In another study, hypoxia, which is also an inducer of oxidative stress [
27], was found to increase the proportion of CD133-positive CSCs in glioma through the activation of the PI3K/AKT and ERK pathways [
6]. A more recent study has shown that the tumor-propagating potential of glioblastoma cells can be enhanced by increasing the expression of a set of transcription factors normally involved in neuronal development, including
POU3F2,
SOX2,
SALL2 and
OLIG2 [
28]. Epithelial-to-Mesenchymal Transition (EMT), which can be linked to the acquisition of cancer stemness [
29,
30], is also known to be regulated by a number of cellular signaling pathways. For instance, activation of the RAS-MAPK pathway was found to induce EMT and increase the size of the CSC-like cell population [
31]. Chaffer et al. have shown that non-CSCs derived from human breast cancers (CD44
low) can convert to CSC’s (CD44
high), and this conversion is mediated by ZEB1, an important EMT transcription factor [
32]. The same group also demonstrated that TGFβ collaborates with the canonical and non-canonical Wnt signaling pathways to induce EMT [
33].
Studies of acquired cancer stemness require appropriate experimental models. Most of the published studies in this field measured how various experimental manipulations change the size of the CSC populations, which are typically detected based on their expression of specific cell surface markers such as CD133 and CD44 [
12,
20,
34‐
38]. In our studies, we employed a different approach. Specifically, the CSC-like population was defined, detected and isolated based on their responsiveness to a Sox2 reporter. In the literature, we are aware of a number of recent publications in which the CSC populations are also identified based on their differential responsiveness to the same Sox2 reporter used in this current study [
13,
39‐
42]. In addition to the Sox2 reporter, we also found one recent publication in which the CSC population from colorectal cancer was identified based on their differential response to a Wnt reporter [
38]. Similar to our RU/RR model, Wnt reporter responsive cells were found to be more stem-like than unresponsive cells, and a very small subset of reporter unresponsive cells were found to convert into reporter responsive cells upon stimulation with hepatocyte growth factor [
38]. Taken together, we believe that the use of differential reporter responsiveness is a valuable experimental approach to study the acquisition of cancer stemness. In contrast with the use of cell surface markers, we believe that this experimental approach has certain advantages, as it provides direct clues to the biochemical mechanisms regulating the cancer stemness phenotype.
From our literature search, the concept of acquired cancer stemness has not yet been described in hematological malignancies, although a good number of CSC markers for these cancers have been proposed, including CD34+/CD38-/CD90−/interleukin 3 receptor(IL-3R)+/human leukocyte antigen (HLA)-DR/CD117- for acute myeloid leukemia (AML), CD34+/CD10-/CD19-/CD133+ for B-cell precursor acute lymphoblastic leukemia (ALL), and CD34+ for chronic myeloid leukemia (CML) [
43,
44]. Regarding ALK+ALCL, we are aware of a publication in which tumor-propagating cells were identified based on their high Hoechst-efflux ability [
45]. These tumor-propagating cells are featured with higher expression levels of NPM-ALK and ATP-binding cassette transporter G2 (ABCG2) as compared to bulk cell populations [
45]. However, there was no evidence of a conversion of non-CSCs to CSCs in this study. Overall, we believe that we have demonstrated the first example of cancer cell plasticity in hematopoietic cancers.
We have previously demonstrated the central role of the Wnt/β-catenin/MYC/Sox2 axis as the defining feature of RR cells in ALK+ALCL [
10]. Specifically, the high level of MYC permits the DNA binding of Sox2, and thus, promotes the transcription of Sox2 downstream target genes. One of the downstream target genes is
WNT2B, which in turn upregulates the Wnt canonical pathway, resulting in an upregulation of MYC [
10]. This positive feedback loop exists in RR cells but not RU cells [
10]. In support of this model, we found that converted RR cells induced by H
2O
2 showed upregulation of this signaling axis. Compared to native RU cells, converted RR cells were found to have a substantially high expression level of Wnt2B, β-catenin, and MYC, as well as enhanced Sox2-SRR2 binding. In contrast, inhibition of β-catenin, MYC, or Sox2 resulted in significantly deceased conversion induced by oxidative stress. Although how exactly oxidative stress stimulates the Wnt/β-catenin/MYC/Sox2 axis is not known, redox oxidative species are known to directly activate the Wnt canonical pathway in HEK293 cells by stabilizing β-catenin and interrupting the interaction between dishevelled (Dvl) and nucleoredoxin (NRX), a strong inhibitor of Wnt/β-catenin signaling [
46]. Further studies are required to determine if these mechanisms indeed underlie the RU to RR conversion in ALK+ALCL cells challenged by oxidative stress.