Tel.: +46-18-611-11-74, Fax: +46-18-55-89-31
Cancer Cell Biology
Forced expression of Sox21 inhibits Sox2 and induces apoptosis in human glioma cells
Article first published online: 9 NOV 2010
Copyright © 2010 UICC
International Journal of Cancer
Volume 129, Issue 1, pages 45–60, 1 July 2011
How to Cite
Ferletta, M., Caglayan, D., Mokvist, L., Jiang, Y., Kastemar, M., Uhrbom, L. and Westermark, B. (2011), Forced expression of Sox21 inhibits Sox2 and induces apoptosis in human glioma cells. Int. J. Cancer, 129: 45–60. doi: 10.1002/ijc.25647
- Issue published online: 26 APR 2011
- Article first published online: 9 NOV 2010
- Accepted manuscript online: 7 SEP 2010 10:20AM EST
- Manuscript Accepted: 10 AUG 2010
- Manuscript Received: 7 JAN 2010
- Swedish Cancer Foundation
- The Children's Cancer Foundation
- Åke Wibergs Foundation
- Magnus Bergvalls Foundation
- Top of page
- Material and Methods
- Supporting Information
Numerous studies support a role for Sox2 to keep stem cells and progenitor cells in an immature and proliferative state. Coexpression of Sox2 and GFAP has been found in regions of the adult brain where neural stem cells are present and in human glioma cells. In our study, we have investigated the roles of Sox2 and its counteracting partner Sox21 in human glioma cells. We show for the first time that Sox21 is expressed in both primary glioblastoma and in human glioma cell lines. We found that coexpression of Sox2, GFAP and Sox21 was mutually exclusive with expression of fibronectin. Our result suggests that glioma consists of at least two different cell populations: Sox2+/GFAP+/Sox21+/FN− and Sox2−/GFAP−/Sox21−/FN+. Reduction of Sox2 expression by using siRNA against Sox2 or by overexpressing Sox21 using a tetracycline-regulated expression system (Tet-on) caused decreased GFAP expression and a reduction in cell number due to induction of apoptosis. We suggest that Sox21 can negatively regulate Sox2 in glioma. Our findings imply that Sox2 and Sox21 may be interesting targets for the development of novel glioma therapy.
Glioblastoma multiforme (GBM) is the most malignant and common form of brain tumor in adults with a mean survival time of <1 year. Recent data suggest that initiation and progression of gliomas are driven by a subpopulation of tumor cells, so called cancer stem cells or brain tumor initiating cells (BTICs), which are clonogenic and able to regenerate the original tumor.1 Cancer stem cells are defined by the ability to unlimited self-renewal, the capacity to initiate and drive tumor development, and to differentiate into different cell lineages. Cancer stem cells have been identified in a number of studies of brain tumors2–4 but exactly which cell is initially transformed remains unclear. Like normal stem cells, also cancer stem cells may have a slower division rate that makes them more resistant to treatment. In brain tumors, cancer stem cells are thus able to resist chemotherapy better than the tumor bulk.5–6
The transcription factor Sox2 is expressed in embryonic stem cells and in stem cells and progenitor cells such as neural stem cells during development and adulthood.7 In cells of the subventricular zone of the lateral ventricles of fetal and adult brain and in the subgranular layer of the adult hippocampus, Sox2 is coexpressed with glial fibrillary acidic protein (GFAP).8, 9 These cells are believed to function as neural stem cells7, 10, 11 and are located in the same areas where brain tumors are thought to arise. When the expression of Sox2 is diminished in these areas, the proliferation of precursor cells and generation of new neurons in adult mice decrease, and the number of GFAP/nestin positive cells is reduced. This suggests that Sox2 is an essential factor for keeping the neural stem cells in a proliferative and undifferentiated state. Further, it has been shown that downregulation of Sox2 blocks the proliferation of multipotent neural stem-like cells (NSLC) and induces neuronal differentiation.12 Moreover, reactivation of Sox2 in O-2A progenitor cells (oligodendrocyte-type-2-astrocyte progenitor cells) converts the cells to NSLC.
Sox2 belongs to the SoxB1 subfamily of the Sox super-family, which comprises Sox1, Sox2 and Sox3. In the chicken embryo, suppression of Sox1-3 promotes neuronal differentiation and overexpression of Sox1-3 completely blocks premature differentiation.13 The SoxB2 subfamily contains the Sox14 and Sox21 proteins, which have a repression domain instead of the activating domain found in the SoxB1 subfamily.14 Sox21 is often coexpressed with Sox2 in chicken, and it has been suggested that a proper balance of Sox21 and Sox2 expression is necessary for the regulation of the target gene.14 In addition, forced expression of Sox21 promotes differentiation of neural cells of the chicken neural tube by reducing the effect of Sox3. In this case, the balance of Sox21 and Sox1-3 decides whether the neural cells should differentiate or stay in a progenitor state.15
We have previously shown that Sox2 is expressed in mouse gliomas induced by PDGFB-encoding retroviruses.16 In addition to Sox2, GFAP, PDGF receptor-α (PDGFRα), NG2 and Sox10 and to some extent nestin are all expressed in the tumors.16 In human, Sox2 has been reported to be expressed in different kinds of pediatric brain tumors such as astrocytomas and high-grade gliomas as well as in undifferentiated and differentiated neurospheres from these tumors.17 Further, Sox2 and GFAP are coexpressed in glial tumors of astroglial, oligodendroglial and ependymal lineages.9
Whereas there is increasing information on Sox2 in gliomas, considerably less is known about its counteracting partner Sox21. The objective of our study was to analyze the presence of Sox21 in glioma cells and study its potential as a Sox2 repressor using human glioma cell lines as a model system. We show for the first time that Sox21 is expressed in brain tumor tissue. Glioma cell lines could be divided into two distinct groups: Sox2+/GFAP+/Sox21+/fibronectin (FN)− and Sox2−/GFAP−/Sox21−/FN+. Upregulation of Sox21 using a tetracycline inducible expression system reduced the expression of Sox2 and GFAP. Further, decreased expression of Sox2 in glioma cells, afforded by Sox21 overexpression or addition of siRNA against Sox2, resulted in reduced cell proliferation through induction of apoptosis.
Material and Methods
- Top of page
- Material and Methods
- Supporting Information
The following human glioma cell lines were used: U-2987 MG, U-343 MG-a Cl2.6, U-87 MG, U-105 MG, U-118 MG, U-138 MG, U-251 MGsp, U-343 MG, U-373 MG, U-1231 MG and U-1242 MG. All lines were of glioblastoma origin and established in our laboratory. For routine culture, cells were grown in Eagle's MEM supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine and 100 units/ml penicillin.
Human GBM BTICs culture
Fresh tumor samples were obtained from adult patients during the operative procedure. The tumors were graded at the Uppsala University Hospital by a neuropathologist according to World Health Organization (WHO) guidelines. The tumor collection was performed in accordance with the human ethical permission. In brief, the tumor tissues were minced into 1 mm × 1 mm pieces and transferred to a tube containing Accutase (eBioscience/Innovative Cell Technologies, San Diego, CA). The tissues were dissociated at 37°C for 45 min and triturated by using 1,000 μl pipette 20 times every 15 min. The cells were centrifuged at 1,000 rpm and resuspended in appropriate volume of DMEM/F12 medium. The cells were filtered through 70-μm cell strainer and resuspended in a medium formulated to sustain the growth of brain tumor initiating cells (BTIC medium,18 containing DMEM/F12 Glutamax (GIBCO/Invitrogen, Carlsbad, CA), 10 mM HEPES (Sigma, Saint Louis, MO), 25 μg/ml insulin, 100 μg/ml transferrin, 20 nM progesterone, 10 μM putrescine, 30 nM selenite, 1% B27 (Invitrogen, Carlsbad, CA), 1% penicillin/streptomycin (Sigma, Saint Louis, MO), 20 ng/ml EGF and FGF2 (Peprotech, Rocky Hill, NJ). The cells were plated at a concentration of 1 × 106 cells per milliliter. After primary sphere formation, the spheres were seeded on dishes coated with 1:10 diluted ECM gel (Sigma, Saint Louis, MO) and cultured as adherent cells as described.18
Immunohistochemistry of human brain tissue
Human brain tumor tissue sections were obtained from patients diagnosed by a clinical neuropathologist according to WHO guidelines at the Department of Genetics and Pathology, Uppsala University Hospital, Uppsala, Sweden. The tumor material was collected in accordance with the human ethical permission. Immunohistochemistry was performed on 6-μm sections using Vectastain ABC system and DAB substrate kit according to manufacturer's protocol (Vector Labs, Burlingame, CA). The following primary antibodies were used: polyclonal anti-Sox2 (AB5603; Chemicon/Millipore, Billerica, MA), polyclonal anti-Sox21 (AF3538; R&D systems, Minneapolis, MN) and polyclonal anti-Sox21 (GT15209 Neuromics, Edina, MN). For biotinylated secondary immunoglobulins, we used polyclonal goat anti-rabbit (E0432) and polyclonal rabbit anti-goat (E0466) (DakoCytomation, Glostrup, Denmark). Sections were mounted with Immu-Mount (Shandon, Pittsburgh, PA), and pictures were taken with a Leica bright field microscope.
Adherent BTICs were seeded onto ECM (1:10) coated glass coverslips in 24-well plates at a concentration of 50,000 cells per well. The cells were cultured in BTICs medium or BTICs medium with 10% FBS (Sigma, Saint Louis, MO) for 7 days. Medium were changed once after 3–4 days of culturing.
Expression vector constructs and tetracycline-induced cells
A cDNA construct of chicken Sox21 containing a myc-tag15 was kindly provided by Jonas Muhr, Ludwig Institute for Cancer Research, Karolinska Institute, Sweden. On the protein level, chicken and human protein sequences are identical to at least 89% (NCBI/BLAST/blastx). The transfected and the translated chicken Sox21 protein are identical to the human Sox21 sequence to at least 87% (NCBI/BLAST/blastx). The Sox21-myc cDNA was cloned into a tetracycline-regulated system (T-Rex System, Invitrogen, Carlsbad, CA) as follows. The Sox21-myc cDNA was cut with restriction enzyme ClaI (New England Biolabs, Ipswich, MA) and cloned into the pcDNA 4/TO expression vector (Invitrogen, Carlsbad, CA) cleaved with EcoRV (New England Biolabs, Ipswich, MA). The pcDNA 6/TR vector was transfected into the glioma cell line U-343 MG-a Cl2.6 with Fugene transfection mix according to the manufacturer's protocol (Roche Diagnostics, Bromma, Sweden). Stable pcDNA 6/TR-clones were selected in MEM medium containing 1.5 μg/ml Blasticidin (Invitrogen, Carlsbad, CA). Stably transfected Cl2.6-pcDNA 6/TR cells were then transfected with the pcDNA 4/TO-Sox21 vector, and stable cell clones were isolated in MEM containing 1.5 μg/ml Blasticidin and 600 μg/ml Zeocin (Invitrogen, Carlsbad, CA). For induction of Sox21-myc expression in the isolated clones, tetracycline (Tet) was added at a concentration of 1 μg/ml. Control cells were treated with absolute ethanol added at the comparative volume, or left untreated.
Transfection with siRNA
siRNA-mediated inhibition of Sox2 was preformed according to the manufacturer's protocol (Santa Cruz Biotechnology, Santa Cruz, CA). Two hundred thousand cells were seeded in 6-well dishes, and after 24 hr, 8 μl of 10 μM Sox2 siRNA was mixed with 8 μl of transfection reagents and transferred to cells in transfection medium at a final concentration of 0.08 μM. The medium was changed after 12 hr of incubation. Cells were harvested for analysis 48 hr thereafter if not otherwise stated. siRNA against Sox2 and control siRNA A were obtained from (Santa Cruz Biotechnology, Santa Cruz, CA). The same results were obtained with siRNA against Sox2 from (Sigma, Saint Louis, MO) (not shown).
Cells were grown on coverslips, fixed with 4% paraformaldehyde and washed in PBS. Cells were blocked in 5% FBS and 0.05% Triton X-100 (for permeabilized cells) in PBS for 1 hr. Sections were then incubated with different antibodies at room temperature for 1 hr followed by incubation with conjugated secondary antibodies. Nuclei were stained with DAPI (Sigma, Saint Louis, MO). The following primary antibodies were used: polyclonal anti-Sox2 (AB5603; Chemicon/Millipore, Billerica, MA), polyclonal anti-Sox21 (AF3538; R&D systems, Minneapolis, MN), monoclonal anti-GFAP (MAB3402; Chemicon/Millipore, Billerica, MA), polyclonal anti-GFAP (Z 0334, DakoCytomation, Glostrup, Denmark) and polyclonal anti-fibronectin (MAB1937; Chemicon/Millipore, Billerica, MA). Secondary antibodies used were: donkey anti-mouse Alexa 488, donkey anti-mouse Alexa 555, donkey anti-rabbit Alexa 488, donkey anti-rabbit Alexa 555 and chicken anti-goat Alexa 488 (Invitrogen, Carlsbad, CA). Pictures were taken with Zeiss 510 confocal microscope and processed with the LSM Image Browser program (Zeiss, Oberkochen, Germany).
Western blot analysis
Cells were grown for 2, 4, 6, 8 and 10 days, if not otherwise stated, in the presence of tetracycline or ethanol or left untreated, and lysed in a buffer containing 20 mM Tris-HCl pH 8.0, 138 mM NaCl, 2 mM EDTA pH 7.0, 1% NP40, 0.5% deoxycholic acid, 35 ng/ml phenylmethylsulfonyl fluoride, 0.1% trasylol, 1 mM Na3VO4, 10 mM NaF and 1 mM ZnCl2. BSA protein assay (Pierce/Thermo Scientific, Rockford, IL) was used to determine protein concentrations according to the manufacturer's protocol. Cell lysates were analyzed on 4–12% Bis–Tris SDS-PAGE gels and electrophoresis was performed according to the manufacturer's protocol (Invitrogen, Carlsbad, CA). Proteins were transferred to a nitrocellulose membrane (Hybond ECL, GE Healthcare, Uppsala, Sweden). The membrane was blocked in 5% BSA (Sigma, Saint Louis, MO), 5% milk (BioRad) or a combination of BSA and milk depending on the antibody used. Immunoblots were developed with Super Signal substrate solution (Pierce/Thermo Scientific, Rockford, IL). The following antibodies were used for Western blot: polyclonal anti-Sox2 (AB5603; Chemicon/Millipore, Billerica, MA), polyclonal anti-Sox21 (AF3538; R&D systems, Minneapolis, MN), monoclonal anti-GFAP (MAB3402; Chemicon/Millipore, Billerica, MA), monoclonal anti-c-myc (9E10:sc-40; Santa Cruz Biotechnology, Santa Cruz, CA), monoclonal anti-neuronal class III β-tubulin (Tuj1) (MMS-435P; Covance, Princeton, NJ), monoclonal anti-CNPase (C5922; Sigma, Saint Louis, MO and SMI-91R; Covance, Princeton, NJ) and monoclonal anti-β-actin (A5441; Sigma, Saint Louis, MO). Western blots were quantified by Fujifilm intelligent Dark Box II/LAS-1000 (Fuji).
Quantitative real-time PCR
Total RNA was isolated from cells treated with siRNA (see above) using the mirVANA PARIS™ kit from Ambion (Applied Biosystems, Austin, TX) according to the manufacturer's protocol. The RNA was treated with DNase (Promega, Madison, WI) before cDNA was synthesized using reverse transcriptase (New England Biolabs, Ipswich, MA) and random hexamers (Invitrogen). Quantitative real-time PCR (qRT-PCR) was performed with Power Syber green Ambion (Applied Biosystems, Austin, TX) according to the manufacturer's recommendations with 0.2 μM of each primer on a Stratagene Mx3000P real-time PCR system (AH Diagnostics, Skärholmen, Sweden). Each PCR was run in triplicates with a 10-min denaturation at 95°C, followed by 45 cycles of denaturation 15 s 95°C, annealing 22 s 55°C and elongation 30 s 72°C. The real-time PCR data were analyzed with the program MxPro-Mx3005P. Transcription levels were normalized against GAPDH levels and mean expression changes were calculated as relative expression levels to cells treated without siRNA. Relative expression changes were calculated with the method.19 Equal amplification efficiencies were studied by calculating ΔCT values from serial dilutions of cDNA amplified with all specific primers and the reference primers (GAPDH).
For cells treated with tetracycline and ethanol, total RNA was isolated using TRIzol (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. The RNA was treated with DNase and qRT-PCR was run in the same way as for the siRNA experiments, with the exception that the annealing temperature was 58°C. The analysis of the results was performed as for the siRNA experiments.
The following primers were used: Sox2: forward primer: GCTCGCAGACCTACATGAAC, reverse: GGGAGGAAGAGGTAACCACA; Sox21: forward primer: CCACTCGCT TGGATTTCTGACACA, reverse: TCGACTCAAACTTAGGG CAACGA; GFAP: forward primer: GCAGAAGCTCCAGG ATGAAACCAA, reverse: GCGACTCAATCTTCCTCTCCAG AT; PDGFRα: forward primer: TGGTTGAAGGAACAGCC TATGG, reverse: TGGCCGTGGGTTTTAGCAT; β-actin: forward primer: GGACTTCGAGCAAGAGATGG, reverse: AGC ACTGTGTTGGCGTACAG; GAPDH: forward primer: ACATCAAGAAGGTGGTGAAGCAGG, reverse: TGTCGCTGTTGA AGTCAGAGGAGA.
For the cell proliferation assay, 20,000 cells were seeded per 35-mm dish, and total number of cells was counted on days 1, 3, 5, 7, 9, 12 and 14 using a Coulter counter (Coulter Electronics, Buckinghamshire, England, UK). Cells were treated daily with tetracycline (1 μg/ml), control cells were treated with absolute ethanol added at the same volume, or left untreated.
Ten thousand cells were seeded per 35-mm dish treated with tetracycline as described above and harvested on days 6, 9, 12 and 14. The annexin binding buffer and propidium iodine were added to the cells (106 cells per milliliter), and the assay was performed according to the manufacturer's protocol (Invitrogen, Carlsbad, CA) and analyzed on a BD (Franklin Lakes, NJ) LSRII (special order system).
For the expression analyses, the in silico transcriptomics (IST) database were used. The database consists of data from Affymetrix gene expression array experiments (the database contains five different array generations). mRNA expression levels of different genes can be compared within the database. IST comprises normal human tissue, cancer tissue and other diseases20 (www.genesapiens.com). We have performed gene expression plots across the entire tissue range for Sox2, Sox21, GFAP and fibronectin, and coexpression plots of Sox2–Sox21 and Sox2–GFAP.
- Top of page
- Material and Methods
- Supporting Information
Sox2, GFAP and Sox21 are coexpressed in glioma cell lines
A number of human glioma cell lines were analyzed for expression of the stem cell factor Sox2 and the astrocytic/neural stem cell marker GFAP. The results were striking; of 11 analyzed glioma cell lines five were positive for nuclear Sox2 expression as well as GFAP expression (Table 1). The expression of Sox2 and GFAP was not always found in the same individual cell but coexpressed in the same cell line (Fig. 1a). Cell lines that were negative for Sox2 expression were consistently negative for GFAP expression. In addition, expression of GFAP and Sox2 was mutually exclusive with the expression of fibronectin (FN)21 with one exception (Table 1). In U-2987 MG cultures, a few cells stained positive for fibronectin whereas the majority were Sox2 and/or GFAP positive (Fig. 1a). We continued to investigate the expression of Sox21, which is found to be coexpressed with Sox2 in chicken.14 Sox21 was expressed in the same cell lines as Sox2 and GFAP with one exception, viz. U-251 MGsp, which was negative for Sox21 expression but expressed both Sox2 and GFAP (Fig. 1b and Table 1). Apart from the exceptions mentioned, the human glioma cell lines could be divided into two distinct groups, Sox2+/GFAP+/Sox21+/FN− and Sox2−/GFAP−/Sox21−/FN+.
Sox2 and Sox21 is expressed in human glioma tissue as well as in primary glioma cell cultures
To corroborate our findings on established glioma cell lines, we analyzed the expression of Sox2 in primary brain tumor tissue by immunohistochemistry. Sox2 was present in both adult and childhood brain tumors and found to be expressed in GBM, astrocytoma Grades I and II, oligodendroglioma, as well as in medulloblastoma, PNET and ependymoma (Fig. 2). This finding confirms the data of Phi et al. that Sox2 is expressed in different kind of brain tumors with one exception; in contrast to the study of Phi et al., we found Sox2 expression in medulloblastomas thus supporting the data of Hemmati et al.9, 17
We then analyzed the expression of Sox21 in the same tumor samples as above. Sox21 was expressed in different kinds of brain tumors, such as astrocytoma Grades I and II, oligodendroglioma, GBM, medulloblastoma, PNET and ependymoma, and the expression was found in both adult and childhood brain tumors (Fig. 2). Adjacent sections were used to compare the expression of Sox2 and Sox21. Sox2 and Sox21 were apparently expressed in the same cell (Fig. 2), but in a number of tumor samples Sox2 was expressed by a larger number of cells compared to Sox21.
Next, expression of Sox2, GFAP, Sox21 and fibronectin was studied by immunofluorescence in a primary glioblastoma cell culture, U-3001 MG, grown under conditions known to sustain cells with tumor initiating capacity.18 Primary U-3001 MG cells had a high expression of Sox2 and also expressed GFAP and Sox21 (Fig. 3) whereas only very few cells stained positive for fibronectin. Sox2 and Sox21 were often coexpressed by the same cell confirming the results of the tumor tissue samples and the stainings of the glioma cell lines (Fig. 3). Interestingly, GFAP and fibronectin were never found to be expressed in the same cell (Fig. 3), supporting our findings in the glioma cell lines. Next, we analyzed the expression of Sox2, Sox21 and GFAP in primary U-3001 MG cells grown in the presence of 10% FBS to induce differentiation. After 7 days in 10% FBS, the cells appeared flat with an increased number of extensions, and the expression of both Sox2 and Sox21 was markedly reduced. In addition, the number of cells expressing GFAP was increased indicating that the cells had undergone astrocytic differentiation. A striking change was the increased number of fibronectin positive cells, although no cells were found to coexpress fibronectin and GFAP. The staining pattern of primary U-3001 MG cells thus supported the idea that Sox2 is expressed in glioma initiating cells and that Sox21 is coexpressed with Sox2. The staining pattern of primary U-3001 MG cells further demonstrated that GBM cells can be divided into at least two distinct groups, Sox2+/GFAP+/Sox21+/FN− and Sox2−/GFAP−/Sox21−/FN+.
Expression levels of Sox2 and Sox21 are correlated in human glioma
We next asked the question how the genes encoding Sox2, Sox21, GFAP and fibronectin are expressed in glioma and other human cancer tissues. For this purpose, we used the IST database, by which it is possible to evaluate and compare mRNA expression levels of human genes (www.genesapiens.com).20 In healthy tissues, Sox2 is mainly expressed in the central nervous system (CNS) and in the respiratory system. In cancer samples, Sox2 is expressed at high levels in glioma and lung cancer but to a lower extent also in head and neck cancer, testicular cancer and cervical cancer (Supporting Information Figure 1A). The expression profile of Sox2 mRNA fits perfectly well with our results from a human normal and tumor tissue microarray (not shown). The IST-expression profile of Sox21 shows specific expression in healthy peripheral nervous system (PNS), in circulating reticulocytes and in muscle. Further, Sox21 is expressed in gliomas (Supporting Information Figure 1B). GFAP is as expected expressed at high levels in CNS and PNS and further at high levels in glioma (Supporting Information Figure 2A). Fibronectin, on the other hand, shows PNS-specific expression but with no significant expression in glioma. Fibronectin is expressed in lung, thyroid and bladder cancer as well as in mesothelioma (Supporting Information Figure 2B). The expression profiles indicate that Sox2, Sox21 and GFAP are interlinked in human glioma and in the normal nervous system. Correlation plots of the mRNA expression of the genes in glioma were constructed using the IST database20 (www.genesapiens.com). As shown in Figure 4a, the expression of Sox2 and Sox21 is correlated, similarly to our findings in primary U-3001 MG cells and glioma cell lines. No clear correlation was found between the expression levels of Sox2 and GFAP. Rather, all the glioma samples express both GFAP and Sox2 but to a varying extent (Fig. 4b).
Sox2 regulates expression of GFAP and Sox21
To investigate the functional role of Sox2 in the Sox2+/GFAP+/Sox21+/FN− glioma cells, we suppressed Sox2 expression using siRNA in U-2987 MG, U-343 MG-a Cl2.6, U-373 MG, U-1231 MG cells, using U-343 MG cells as a negative control (Fig. 5). siSox2 treatment of glioma cells led to a marked reduction of GFAP (Figs. 5a and 5b) and Sox21 (Fig. 5a). The effects of siSox2 treatment were validated by qRT-PCR. RNA was collected from U-2987 MG cells treated with or without siRNA against Sox2 and the expression of Sox2, GFAP, Sox21 and PDGFRα was analyzed. The results are presented as relative expression of mock-treated cells (w/o) compared to siSox2 treated cells and normalized to expression of the housekeeping gene GAPDH (glyceraldehydes-3-phosphate dehydrogenase). β-Actin was used as an additional control whose relative expression remained unchanged. Sox2, GFAP, Sox21 and PDGFRα were clearly downregulated at the RNA level confirming the Western blot data (Fig. 6). Our data indicate that downregulation of Sox2 with siRNA is in addition reducing the expression of GFAP and Sox21, suggesting that Sox2 regulates the expression of both GFAP and Sox21.
Overexpression of Sox21 suppresses Sox2 expression
To further investigate the role of Sox2 in glioma cells, we used Sox21 as a tool to regulate Sox2 in Sox2+/GFAP+/Sox21+/FN− cells. In chicken embryos, Sox21 has been reported to counteract the activity of Sox2 and promote differentiation.15 We used a tetracycline-inducible system where Sox21 expression could efficiently be upregulated in U-343 MG-a Cl2.6 cells. Three inducible cell clones were selected and denominated Cl2.6 clone A1-Sox21, Cl2.6 clone A2-Sox21 and Cl2.6 clone A3-Sox21. Cl2.6 clone B1-empty is a negative control containing the primary vector without the Sox21 gene. In the following experiments, the different cell clones were treated with tetracycline (Tet) or ethanol daily up to 10 or 14 days, or left untreated. All Sox21 clones displayed a Tet-inducible expression of Sox21. The expression of Sox21 was detected both with an antibody against Sox21 (detecting induced as well as endogenous Sox21 expression) and an antibody detecting the myc-tagged Sox21 (induced Sox21) (Fig. 7). Only a faint band was detected in the control clone, Cl2.6 clone B1-empty, showing endogenous expression of Sox21 (Fig. 7d). Sox21 overexpression led to a significant downregulation of Sox2 in two of the inducible clones, Cl2.6 clone A1-Sox21 and Cl2.6 A2-Sox21 (Figs. 7a and 7b). Interestingly, Cl2.6 clone A3-Sox21 showed very low levels of Sox2 expression, if any, even under noninducing conditions (Fig. 7c). Since Cl2.6 clone A3-Sox21 express no or very low levels of Sox2 this clone could be used as a control for Sox2 independent effects of induced Sox21 expression.
We continued to examine whether the induced overexpression of Sox21 regulates Sox2 on the RNA level. As determined by qRT-PCR, all cell clones expressed Sox2 mRNA, except for Cl2.6 clone A3-Sox21 (Supporting Information Figure 3). The amount of Sox2 mRNA in Cl2.6 clone A3-Sox21 cells was <10% of that of Cl2.6 clone A2-Sox21 cells, in agreement with the low levels of Sox2 protein. However, when Sox21 was overexpressed in the inducible clones there was no significant downregulation of the mRNA levels of Sox2 (Supporting Information Figure 3), despite the reduced amount of Sox2 protein. The qRT-PCR results therefore indicate that Sox21 in this Sox21-inducible cell system does not regulate Sox2 at the level of mRNA expression.
Induced Sox21 expression inhibits cell proliferation
To test our hypothesis that Sox2 expression is indispensible for keeping the cells in an immature proliferating state, we next investigated the effect of Sox2 downregulation on the proliferation rate in Cl2.6-Sox21 inducible clones. Cell proliferation after 14 days of Sox21 induction was markedly inhibited in clones where Sox2 expression was downregulated (Figs. 8a and 8b). In contrast, the proliferation of clone A3-Sox21 cells, with no endogenous expression of Sox2, or clone B1-empty cells, was unaffected. This implies that Sox21 overexpression leads to a retarded cell proliferation and that this effect may be caused by Sox2 downregulation in Sox2 positive cells.
To analyze this further, we next investigated the effect of siSox2 on cell proliferation. Forty-eight hours after siSox2 transfection, a clear reduction in cell number was detected in both of the tested Sox2 positive cell lines, U-2987 MG and U-343 MG-a Cl2.6. In U-2987 MG cells, the effect was maintained for at least 192 hr (Fig. 9a). In U-343 MG-a Cl2.6 cells, the effect lasted about 120 hr, but at later time points the proliferation rate was similar to that of mock-transfected cells (Fig. 9b). The inhibition of cell proliferation was higher in U-2987 MG where the suppression of Sox2 by siRNA was more efficient compared to U-343 MG-a Cl2.6 cells (Fig. 5). Our results support the notion that Sox2 is involved in the regulation of cell proliferation in glioma cells.22
Since Sox2 keeps glioma cells in a proliferative state and Sox21 counteracts the effect of Sox2, we analyzed if the induction of Sox21 would lead to a different expression of neural/glial markers in the glioma cells. We found a significant change in GFAP expression in Cl2.6 A2-Sox21 where the GFAP expression was markedly reduced after 6 days of Sox21 induction compared to uninduced cells (Fig. 10a). The other inducible clones were GFAP negative in the uninduced state (not shown), but Cl2.6 clone B1-empty was GFAP positive and remained unaffected after addition of tetracycline (Fig. 10b). The downregulation of GFAP Cl2.6 A2-Sox21 was confirmed by qRT-PCR. The expression of GFAP mRNA was reduced after 4 days of Sox21-induction (Supporting Information Figure 3). Both CNPase and Tuj1 were expressed by all clones with no significant changes in expression upon Sox21 induction (not shown). These observations suggest that upon Sox2 downregulation by Sox21, the cell proliferation rate is strongly reduced but the cells do not show any obvious signs of differentiation.
Upregulation of Sox21 induces apoptosis
Next, we asked the question whether the reduced cell proliferation rate was due to a growth arrest or apoptosis. Cell cycle analysis by flow cytometry after staining with propidium iodide showed no significant difference between Sox21 induced cells, ethanol-treated or untreated cells (not shown). An induction of apoptosis was evident in all three Sox21 inducible clones (Figs. 11a–11c), i.e., the number of annexin V positive cells was increased after incubation in tetracycline-containing medium compared to the effect of ethanol. The fraction of dead cells was significantly higher as well as the fraction of cell debris in all three Sox21 inducible clones compared to ethanol-treated cells. This resulted in a significant reduction of the number of living cells in the Sox21 inducible clones compared to ethanol-treated cells (Figs. 11a–11c). Cl2.6 clone A3-Sox21 that had no endogenous expression of Sox2 was also affected but the percentage of living cells was considerably higher in this clone compared to the other two Sox21 inducible clones, in the same range as for Cl2.6 clone B1-empty (Figs. 11c and 11d). Apoptosis was not induced to the same extent in the control clone (Cl2.6 clone B1-empty) and the distribution of cells in the different fractions, tetracycline- and ethanol-treated cells, did not differ (Fig. 11d). An upregulation of cleaved caspase-3 in Cl2.6 clone A2-Sox21 was detected on Western blot after Sox21 induction (not shown), supporting the idea that an upregulation of Sox21 induces apoptosis. Taken together, our results indicate that an upregulation of Sox21 in glioma cells suppresses Sox2 expression and induces apoptosis, which results in reduced cell proliferation.
- Top of page
- Material and Methods
- Supporting Information
A striking finding of the present investigation is that human glioma cell cultures can be divided into two distinct groups: Sox2+/GFAP+/Sox21+/FN− and Sox2−/GFAP−/Sox21−/FN+ (Table 1). The coexpression of Sox2 and GFAP in glioma cells is in line with the current idea that an adult neural subventricular zone stem cell could be the initiating cell in the genesis of glioblastoma. The derivation of the Sox2−/GFAP−/Sox21−/FN+ glioma cell is more enigmatic and remains to be revealed. In fact, the majority of published glioma cell lines display this mesenchymal-like phenotype.23 Interestingly, recent studies have shown that glioblastoma stem cells may give rise to a progeny with mesenchymal-like characteristics.24–26 The Sox2−/GFAP−/Sox21−/FN+ glioma cell lines may be representatives of this lineage and selected for in culture by its propensity to grow adherently in serum-containing medium.
Sox2 and Sox21 have been shown to be coexpressed in chicken brain where a proper balance of Sox21 and Sox2 expression has been proposed to be necessary for the regulation of target genes in neural stem cells.14 We report here that human glioma cells of the Sox2+/GFAP+/FN− group express Sox21. Both in the primary U-3001 MG cells and in the glioma cell lines, Sox2 and Sox21 are found to be expressed by the same cell. This is further supported by the immunohistochemistry stainings of brain tumor sections. The in silico expression analyses of human glioma samples confirm that Sox21 is expressed in glioma and that there is a correlation between the expression of Sox2 and Sox21. Both the immunofluorescense staining of the cells and the expression analyses suggest that the expression of Sox2 and Sox21 is interdependent such that a high expression of Sox2 correlates with a high expression of Sox21.
Downregulation of Sox2 expression by siRNA reduces the expression of GFAP and Sox21 at the protein level as well as at the mRNA level. Moreover, PDGFRα is also downregulated at mRNA levels indicting a correlation between Sox2 and PDGFRα. We have previously shown that Sox2 is widely expressed together with PDGFRα in PDGFB induced gliomas16 and PDGFRα is expressed in neural stem cells in the subventricular zone.27 Amplification of the PDGFRα gene is a common event in high-grade gliomas.28 Incubation of Sox2 expressing cells with siSox2 reduced cell proliferation, which indicates that Sox2 is involved in the growth control of glioma cells. Similarly, Gangemi et al. have shown that when Sox2 is silenced with miRNA the cell proliferation rate is evidently reduced in primary glioblastoma tumor initiating cells.22
The downregulation of GFAP as an effect of Sox2 targeted siRNA indicates a, direct or indirect, functional connection between Sox2 and GFAP. Cavallaro et al. also show that Sox2 can bind the GFAP promoter and in neural cells that have already entered the differentiation program an overexpression of Sox2 reduces GFAP expression,29 whereas in our study, downregulation of Sox2 reduces GFAP expression. This could indicate that the glioma cells used in our study are of more immature character as those used in the study by Ferri et al. where GFAP and nestin positive cells are reduced in the absence of Sox2.7 Also in the neural tube of chicken Sox2 expression prevent neural differentiation and maintain the cells in a proliferative stage.13 The details of GFAP regulation in glioma cells remains to be elucidated.
To further investigate the role of Sox2 in glioma cells, we used Sox21 as a potential negative regulator of Sox2. Sox21 has been shown to be coexpressed with Sox2 in chicken brain, and a forced expression of Sox21 results in neuronal differentiation.14, 15 Indeed, upregulation of Sox21 results in inhibition of Sox2 and GFAP expression and reduced cell proliferation in Sox2 positive clones. The proliferation of the Sox2 negative Cl2.6 clone A3-Sox21 is unaffected by Sox21 overexpression, suggesting that the effect of Sox21 on proliferation is indirect and mediated by Sox2 downregulation. Sox2 may therefore be essential for keeping the tumor cells proliferative and thus mimicking its role in normal stem cells and progenitor cells.12, 30
There were no changes in the cell cycle distribution (not shown), or in the expression of differentiation markers in Sox21 overexpressing cells. An increase of apoptotic, dead cells and debris is noticed in the three Sox21-induced clones compared to ethanol-treated cells (Fig. 11). There are also signs of apoptosis in the Sox2 negative clone (Cl2.6 clone A3-Sox21), but the fraction of living cells is higher compared to the other Sox21 inducible clones. These results suggest that the reduced proliferation rate is at least partially a consequence of apoptosis, which is induced by Sox2 downregulation.
In addition to the finding that forced expression of Sox21 downregulated the expression of Sox2, we also found that exposure to siSox2 downregulated Sox21, both at the protein and at the mRNA level. This finding suggests a regulatory circuit where Sox2 positively regulates Sox21 whereas Sox21 negatively regulates Sox2. Since Sox2 and Sox21 are coexpressed in glioma cells there might be a balance between Sox2 and Sox21 such that the levels of Sox21 are not high enough to effectively repress Sox2. If Sox2 expression is obligatory for the progressive growth of glioma cells in vivo as our studies indicate, Sox2 inhibitors, Sox21 activators, or downstream or upstream effectors may be interesting targets for glioma therapy. Indeed, recent data have shown that inhibition of the TGFβ receptor pathway leads to a reduction of Sox2 levels and growth inhibition of glioma initiating cells.31
- Top of page
- Material and Methods
- Supporting Information
We thank Dr. Jonas Muhr at the Ludwig Institute for Cancer Research, Karolinska Institute, Sweden for providing the Sox21 cDNA. We also thank Emma Bergfelt for skilful technical assistance.
- Top of page
- Material and Methods
- Supporting Information
- 21Human Glioma Cell Lines. In: HayRJ, ParkJG, GazdarA, eds. Atlas of Human Tumor Cell Lines. New York: Academic Press, Inc., 1994. 17–42., .
- 23Mechanisms of altered growth control. Growth factors. In: BignerDD, McLendonRE, BrunerJM, eds. Russel and Rubinstein's pathology of tumors of the nervous system, vol. 1. London: Arnold, 1998. 83–116., .
- Top of page
- Material and Methods
- Supporting Information
Additional Supporting Information may be found in the online version of this article.
|IJC_25647_sm_suppfig-S1.tif||30368K||Supplementary figure 1. Detailed human expression profiles of Sox2 and Sox21. The insilico transcriptomics (IST) database have been used to compare the mRNA expression profiles of A, Sox2, B, Sox21 20 (www.genesapiens.com). The expression profiles are divided into healthy human tissues, cancers and other diseases as indicated. Specific expression is highlighted in tissue-specific colors as indicated.|
|IJC_25647_sm_suppfig-S2.tif||30051K||Supplementary figure 2. Detailed human expression profiles of GFAP and fibronectin. The insilico transcriptomics (IST) database have been used to compare the mRNA expression profiles of A, GFAP and B, fibronectin 20 (www.genesapiens.com). The expression profiles are divided into healthy human tissues, cancers and other diseases as indicated. Specific expression is highlighted in tissue-specific colors as indicated.|
|IJC_25647_sm_suppfig-S3.tif||27844K||Supplementary figure 3. Induced Sox21 expression did not downregulate Sox2 on the mRNA level but the GFAP mRNA was reduced. The four U-343 MG-a Cl2.6 clones were treated with tetracycline (Tet) or ethanol (EtOH) up until 12 days. RNA was prepared and qRT-PCR was performed. A, Show Cl2.6 clone A1-Sox21, Cl2.6 clone A2-Sox21 and Cl2.6 clone A3-Sox21 respectively and the effect of induced Sox21 on Sox2 mRNA level. Cl2.6 clone B1-empty is the negative control clone with no induction of Sox21. B, Cl2.6 clone A2-Sox21 show the effect of induced Sox21 on GFAP mRNA level, Cl2.6 clone B1-empty is used as a negative control clone without Sox21 cDNA. Shown is one representative experiment out of 2. The values are represented as mean +/−s.d. (n=3 for each experiment).|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.