|Home | About | Journals | Submit | Contact Us | Français|
Emerging evidence suggests a decline of ERβ expression in various peripheral cancers. ERβ has been proposed as a cancer brake that inhibits tumor proliferation. In the current study, we have identified ERβ5 as the predominant isoform of ERβ in human glioma and its expression was significantly increased in human glioma as compared with non-neoplastic brain tissue. Hypoxia and activation of hypoxia inducible factor (HIF) increased ERβ transcription in U87 cells, suggesting elevated ERβ expression in glioma might be induced by the hypoxic stress in the tumor. Over-expression of either ERβ1 or ERβ5 increased PTEN expression and inhibited activation of the PI3K/AKT/mTOR pathway. In addition, ERβ5 inhibited the MAPK/ERK pathway. In U87 cells, ERβ1 and ERβ5 inhibit cell proliferation and reduced cells in the S+G2/M phase. Our findings suggest hypoxia induced ERβ5 expression in glioma as a self-protective mechanism against tumor proliferation and that ERβ5 might serve as a therapeutic target for the treatment of glioma.
Loss of ERβ expression has been suggested as an important step in estrogen-dependent tumor progression (Bardin et al., 2004; Burns and Korach, 2012). In breast cancer, decline of ERβ expression has been repeatedly reported (Bardin et al., 2004;Murphy and Leygue, 2012). High levels of ERβ are associated with lower tumor grade (Jarvinen et al., 2000), longer survival of patients (Honma et al., 2008), and higher sensitivity to tamoxifen treatment (Esslimani-Sahla et al., 2004; Hopp et al., 2004). DNA methylation in the ERβ promoter region has been proposed to contribute to the decline of ERβ expression in tumor cells (Zhao et al., 2003). Both in vitro and in vivo studies indicate that ERβ inhibits proliferation and invasion of breast cancer cells (Lazennec et al., 2001; Paruthiyil et al., 2004). In addition, the anti-proliferative role of ERβ has been demonstrated in hormone-independent cancers of the colon and lung (Hartman et al., 2009; Skov et al., 2008). Different mechanisms have been proposed for the anti-proliferative action of ERβ (Bardin et al., 2004), including inhibition of ERα transcriptional activity (Hall and McDonnell, 1999), reduction of S+G2/M phase (Liu et al., 2002; Strom et al., 2004), and inhibition of HIF1 transcriptional activity (Lim et al., 2011).
At least 5 different isoforms of human ERβ have been identified which have identical N-terminal sequence but diverge from amino acid 469 to the C-terminus (Moore et al., 1998). In vitro analysis has found that each ERβ isoform has distinct transcriptional activity (Leung et al., 2006; Moore et al., 1998). In breast cancer, expression levels and functions of different ERβ isoforms have been studied (Leygue et al., 1999; Omoto et al., 2003; Shaaban et al., 2008). Most studies on ERβ expression in cancer used antibodies that did not discriminate between different ERβ isoforms, and functional analysis of ERβ in cancer has mainly focused on ERβ1. Two recent studies indicated that ERβ expression declined in human glioma as tumor grade increased (Batistatou et al., 2006; Sareddy et al., 2012) and that an ERβ agonist inhibited proliferation of glioblastoma multiforme (GBM) cell lines (Sareddy et al., 2012). However, these studies used only immunohistochemistry to evaluate ERβ expression. It was not clear which isoforms are expressed in human glioma and the distinct function of the each ERβ isoform is unknown. In the present study, we evaluated the expression of ERβ isoforms in human glioma using immunohistochemistry, Western blot, and real time PCR. In addition, the function of ERβ1 and ERβ5 in glioma progression was determined using human GBM cell lines.
We evaluated ERβ expression in human GBM specimens and non-neoplastic brain specimens by Western blot using three different antibodies (H150, 1531 and 3576). With each antibody, we detected an increase of ERβ expression in a human GBM patient as compared with a non-neoplastic brain tissue from female patients (Fig. 1A). The immunoreactive band we detected in Western blots was of a smaller molecular weight than that of full length ERβ1 (59 kDa). We used PCR primers for ERβ isoforms 1, 2, 3, 4 and 5 to determine the expression of each ERβ isoform in human GBM specimens (Moore et al., 1998). In human GBM cDNA from the same patient as in Fig. 1A, only ERβ5 was detected by PCR (Fig. 1B). This was confirmed by PCR using an alternative set of primers for ERβ1, ERβ2, ERβ4 and ERβ5 (Leung et al., 2006) (Supplement Fig. 1C). We evaluated the expression of ERβ isoforms using isoform specific antibodies for ERβ1, ERβ2 and ERβ5. Consistent with the PCR results, higher level of ERβ5 was detected in human GBM by Western blot and immunohistochemistry as compared with non-neoplastic brain tissue (Fig. 1C and D). No positive staining was detected in either non-neoplastic brain tissue or GBM using ERβ1 or ERβ2 specific antibodies with immunohistochemistry. Double immunofluorescence staining demonstrated that ERβ5 positive cancer cells expressed astrocyte marker glial fibrillary acidic protein (GFAP), and that ERβ5 was mainly localized in the nucleus (Fig. 1E).
We further investigated ERβ expression in a different set of non-neoplastic human brain and GBM specimens. Western blot analysis indicated an increase of ERβ5 expression in human GBM from both male and female patients as compared with non-neoplastic brain tissue (Fig. 2A and B). Consistently, real-time PCR demonstrated increased mRNA levels of ERβ5 in GBM tissue from female patients (control: n=3; GBM: n=8) (Fig. 2C). We evaluated ERβ5 expression in different grades (II, III and IV) of human glioma by immunohistochemistry using commercial human brain tissue array samples. Immunohistochemistry scoring indicated a significant increase of ERβ5 expression in gliomas as compared with non-neoplastic brain tissues (Fig. 2D and E). ERβ5 expression was not significantly different between different grades of glioma, although a trend of ERβ5 increase was observed in high grade glioma.
We evaluated the expression of ERβ in human GBM cell lines U87 and A172, which originated from female and male GBM patients, respectively. Using antibodies specific for ERβ1, ERβ2 or ERβ5, we identified the expression of all the three isoforms in both U87 (Fig. 3A) and A172 (Fig. 3B) cells. Expression of ERβ1, ERβ2 and ERβ5 in U87 and A172 was also confirmed by PCR using two different sets of primers (Supplements Fig. 1A and B). All three isoforms were localized in the nucleus as indicated by their co-staining with DAPI. We evaluated expression of ERβ by immunocytochemistry in human primary astrocytes. No positive staining was observed (Supplement Fig. 2A).
Gliomas, like many other solid tumors, are under extensive hypoxic stress given their rapid proliferation (Cruickshank et al., 1994). It has been demonstrated that hypoxia increased ERβ mRNA levels in rat hippocampus neuronal cultures (Heyer et al., 2005). We predicted that hypoxia in glioma might contribute to the increase of ERβ expression. U87 cells were subjected to hypoxic condition (1% O2) for 24 h and the expression of ERβ1, ERβ2, and ERβ5 were evaluated by real-time PCR. We found that hypoxia significantly increased mRNA levels of all three ERβ isoforms (Fig. 4A).
In the promoter region of ERβ, there is an E-box which overlaps with the HIF binding sequence (5′-RCGTG-3′) (Li et al., 2000). We predicted that the hypoxia-induced increase of ERβ transcription is mediated by the activation of HIF. We transfected U87 cells with plasmids P1P2N HIF1α and P1P2N HIF2α to express constitutively active HIF1α or HIF2α, respectively. At 48 h after transfection, RNA was isolated and real-time PCR was conducted to evaluate mRNA levels of ERβ1, ERβ2 and ERβ5. Compared with vector transfection, both P1P2N HIF1α and P1P2N HIF2α transfection significantly increased mRNA levels of all the three ERβ isoforms (Fig. 4B).
Previous studies demonstrated that ERβ1 can increase PTEN expression and suppress activation of the PI3K/AKT pathway (Lindberg et al., 2011). PTEN is an important tumor suppressor and somatic mutation of PTEN is common in human glioma (Wang et al., 1997). To investigate the effect of ERβ on PTEN expression, we cloned ERβ1 and ERβ5 into a pCDNA vector with a flag tag at the N-terminal. The plasmids were transfected into HEK 293 cells, which do not have endogenous ERβ expression. Nuclear localization of ERβ1 and ERβ5 was observed by immunofluorescent staining using anti-flag antibody in the cells with transient ERβ1 or ERβ5 transfection. In addition, the ERβ1/5 localization was not affected by the treatment of 17β-estradiol (E2) (Fig. 5A). In the ERβ1 and ERβ5 stably transfected HEK 293 cell lines, an increase of PTEN expression was observed (Fig. 5B and D). Consistently, we detected decreased phosphorylation of AKT and P70S6K, a component of the mTOR pathway, in both cell lines (Fig. 5B). These data suggest that ERβ1 and ERβ5 could inhibit the PI3K/AKT/mTOR pathway through upregulation of PTEN expression in an estrogen-independent manner. We further tested the effects of ERβ1 and ERβ5 on the activation of the MAPK/ERK pathway. In ERβ5, but not ERβ1, expressing HEK 293 cells, we observed decreased phosphorylation of c-Raf, MEK and ERK in a ligand-independent manner (Fig. 5C).
A recent study indicated that PTEN elevation can increase energy expenditure and mitochondrial oxidative phosphorylation to reduce the Warburg effect of tumor cells (Garcia-Cao et al., 2012). We investigated the oxygen consumption rate of HEK 293 cells with stable expression of either ERβ1 or ERβ5. A significant increase of oxygen consumption rate was observed in ERβ1 and ERβ5 expressing cells as compared with vector control (Fig. 5D).
The effects of ERβ agonist diarylpropionitrile (DPN) on tumor growth have been contradictory. While DPN has been reported to inhibit growth of breast cancer, colon cancer and GBM cells, it has also been demonstrated that DPN could increase proliferation of medulloblastoma cells (Belcher et al., 2009). We treated U87 cells (in phenol red free medium with 10% charcoal-stripped FBS) with E2 or DPN at concentrations of 100 nM. A six-day growth curve indicated no obvious change in the cell growth rate (Fig. 6A). Since our data indicated that ERβ1/5 could inhibit PI3K/AKT/mTOR and MAPK/ERK pathways in a ligand-independent manner, we hypothesize that ERβ1/5 inhibits GBM proliferation in a similar fashion. To determine the effects of ERβ on glioma cell proliferation, we over-expressed ERβ1 or ERβ5 in U87 cells. U87 cells with over expression of either ERβ1 or ERβ5 showed a significantly lower growth rate as compared to vector control cells (p<0.01) (Fig. 6B). Furthermore, ERβ5 expressing U87 cells had an even lower growth rate than ERβ1 expressing cells (p<0.01) (Fig. 6C). Western blot analysis indicated that ERβ5, but not ERβ1, decreased activation of ERK in U87 cells (Fig. 6C). In addition, we observed decreased MMP2 activation in ERβ1 and ERβ5 expressing U87 cells, which was independent of E2 treatment (Fig. 6C).
We evaluated the role of ERβ1 and ERβ5 on cell cycle in U87 cells. Cell cycle analysis indicated that, at 12 h after transfection, ERβ1 or ERβ5 over-expression lead to a reduction of cell number in the proliferative S+G2/M phase (Fig. 6D). At 24 h, a significant reduction of cell number in the S+G2/M phase was found in the ERβ1 or ERβ5 over-expression U87 cells (Fig. 6D).
ERβ has long been shown to be involved in carcinogenesis and cancer progression and recognized as a cancer brake in the ovarian, prostate, and colon cancers (Bardin et al., 2004). The identification of ERβ isoforms has added further complexity to the action of ERβ (Moore et al., 1998). In the current study, we determined the expression of each ERβ isoform in human glioma using immunohistochemistry, Western blot, and real time PCR. We have found that ERβ5 is the main ERβ isoform expressed in human glioma. Consistently, ERβ5 was also identified in two human glioma cell lines, U87 and A172, by PCR and immunohistochemistry. The identified expression of ERβ1, ERβ2 and ERβ4 in U87 and A172 cells might be due to the in vitro culture conditions which could not precisely replicate the glioma cells’ in vivo micro-environment. We found that the expression level of ERβ was low in non-neoplastic brain tissue as indicated by Western blot and PCR. In primary human astrocytes, no obvious positive staining for ERβ was observed by immunocytochemistry. However, in human glioma specimens, we found a significant increase of ERβ5 expression as compared in non-neoplastic brain tissue. A trend of increase of ERβ5 expression was indicated in high grade glioma, although it was not statistically different probably due to the limited sample size. Our results contradict to two recent studies, which reported that ERβ expression declined in human glioma as tumor grade increased (Batistatou et al., 2006; Sareddy et al., 2012). The discrepancy might be due to the different methods between our study and previous studies. In addition, the previous studies did not differentiate each ERβ isoform. The present study argues that future studies should be conducted using ERβ isoform specific antibodies and real-time PCR to further investigate the expression of ERβ isoform in human glioma.
ERβ isoform messenger RNA (mRNA) sequence analysis has identified two different 5′-untranslated regions (5′UTR) composed of two distinct untranslated first exons, indicating that transcription of different human ERβ isoforms occurs from at least two different promoters, namely 0 K and 0 N (Hirata et al., 2001). Further analysis has identified that ERβ5 is regulated exclusively by promoter 0 K, while ERβ1, ERβ2 and ERβ4 are under the control of both promoter 0 K and promoter 0 N (Yuet-Kin Leung, 2005). The reduction of ERβ1, ERβ2, and ERβ4 transcription has been attributed to the methylation of promoter 0 N in prostate, breast, and ovarian cancers, while promoter OK was not methylated (Nojima et al., 2001; Rody et al., 2005; Suzuki et al., 2008; Zhao et al., 2003). The higher level of ERβ5 in glioma demonstrated in our study indicates that the activation of promoter 0 K. While the absence of ERβ1, ERβ2, and ERβ4 expression in glioma suggests that promoter 0 N might be silenced by methylation in glioma. An E-box (5′-CACGTG-3′,−94/−99) has been found in the promoter region of ERβ (Li et al., 2000), which overlaps the HIF binding sequence (5′-RCGTG-3′). We postulated that the increased expression of ERβ5 in glioma was caused by hypoxia that commonly exists in glioma (Evans et al., 2004). Consistently, in U87 cells, hypoxia increased the transcription of ERβ1, ERβ2 and ERβ5. Furthermore, significant increases in ERβ1, ERβ2 and ERβ5 transcription were induced in U87 cells by transfection of constitutively active HIF1α or HIF2α. Both in vitro and in vivo studies have shown that ERβ inhibits cancer. ERβ1 has been most extensively studied. In tamoxifen-treated breast cancer, ERβ5 expression was positively associated with better survival (Davies et al., 2004;Shaaban et al., 2008). However, in prostate cancer, ERβ5 was associated with poor prognosis, and over-expression of ERβ5 promoted cancer cell migration and invasion (Leung et al., 2010). These data suggest that the functions of ERβ5 might be cancer type specific. In vitro analysis indicated that ERβ5 can activate gene transcription without ligand binding (Poola et al., 2005). In the current study, we investigated the effects of ERβ5 on two important oncogenic pathways for glioma: PI3K/AKT/mTOR and MAPK/ERK. We found that ERβ5 inhibited both pathways in HEK 293 cells in ligand-independent manners.
Both ERβ1 and ERβ5 inhibited the PI3K/AKT/mTOR pathway by increasing PTEN expression. An elevated level of PTEN, in vivo, has been shown to increase oxygen consumption and reduce the Warburg effect in tumor cells (Garcia-Cao et al., 2012). Consistently, in ERβ1 or ERβ5 expressing HEK 293 cells, we observed an increased cellular oxygen consumption rate. In breast cancer cells, ERβ over-expression increased PTEN level in a ligand-independent manner (Lindberg et al., 2011). We predicted that up-regulating PTEN expression is a common anti-proliferative mechanism for ERβ. However, in U87 cells, an in frame deletion causes the cells to be devoid of functional PTEN (Furnari et al., 1997). Further studies using GBM cell lines without a PTEN mutation are needed to support our prediction. In the promoter region of PTEN, no estrogen-responsive element (ERE) or ERE-like sequence has been identified (Kamalakaran et al., 2005). Yang et al. proposed that estrogen receptor could bind to the C terminus of PTEN and reduce protein turnover (Yang et al., He). Wickramasinghe et al. (2009) proposed that estradiol bound ER down-regulates miR-21 level and releases suppression of miR-21 on PTEN expression. A recent study indicated that ERβ could bind to the promoter region of PTEN through Sp1 and increase PTEN transcription (Guido et al., 2012). The inhibitory action of ERβ1 and ERβ5 on U87 proliferation identified in the current study also suggests PTEN-independent mechanisms. Indeed, both ERβ1 and ERβ5 decreased MMP2 expression and ERβ5 inhibited the MAPK/ERK pathway. Consistent with previous studies (Liu et al., 2002; Strom et al., 2004), we found that, after over-expression of ERβ1 in U87 cells, cell number at the proliferative fraction S+G2/M phase was significantly reduced. Interestingly, ERβ5 over-expression produced more dramatic reduction of cell numbers in S+G2/M phase. A recent study indicated that ERβ agonist DPN inhibited proliferation of GBM cell lines including U87 cell line (Sareddy et al., 2012). However, we were not able to detect the inhibitory effect of DPN on U87 cell proliferation in the growth curve analysis. Moreover, another previous study demonstrated that DPN increased proliferation of medulloblastoma cells (Belcher et al., 2009). More detailed studies should be done to investigate DPN’s effects on different types of brain tumors before any clinical trial in human is conducted. It has also been demonstrated that a tamoxifen metabolite, endoxifen, can induce ERα/β heterodimer formation to stabilize ERβ protein (Wu et al., 2011). In breast cancer, higher ERβ expression level has been associated with higher sensitivity to tamoxifen treatment (Esslimani-Sahla et al., 2004; Hopp et al., 2004). Clinical trials have indicated that a subgroup of glioma patients responded to tamoxifen treatment (Couldwell et al., 1996; Patel et al., 2012). It would be very interesting to evaluate the expression of ERβ5 in glioma and examine the correlation between ERβ5 expression and the response to tamoxifen treatment.
In summary, our current study demonstrated that ERβ5 is the major ERβ isoform which expression was elevated in human glioma. The increase of ERβ5 might be induced via HIF signaling. In addition, our study indicated that ERβ5 functions as a potent suppressor against glioma progression through multiple mechanisms including suppression of oncogenic PI3K/AKT/mTOR and MAPK/ERK pathways in a ligand independent manner (Fig. 7).
Non-neoplastic human brain tissues (control) and human GBM specimens were collected in Department of Pathology, University of Texas Southwestern Medical Center (Dallas, TX) and Beijing Tiantan Hospital (Beijing, China). Non-neoplastic brain tissues from nine patients and glioma specimens from 22 patients were used for Western blot analysis and cDNA synthesis. Paraffin-embedded GBM specimens from two patients and two non-neoplastic brain specimens were obtained from Beijing Tiantan Hospital. Brain tissue arrays were purchased from US Biomax (Rockville, MD). Immunohistochemistry was done as described previously (Li et al., 2011). After immunohistochemistry of the tissue arrays, 7 non-neoplastic control tissues, 36 grade II glioma, 14 grade III glioma and 6 grade IV glioma were included into quantitative analysis. Images of the tissues on the arrays were captured under the same exposure condition using a Zeiss Microscope (Carl Zeiss). Images were randomly numbered and given to a researcher who scored them blindly. Scoring of the images was conducted as described previously (Harvey et al., 1999). First, a proportion score was assigned to represent the estimated proportion of positively stained cells (0: none; 1: < 1/100; 2: 1/100 to 1/10; 3: 1/10 to 1/3; 4: 1/3 to 2/3; 5: > 2/3); then, an intensity score was assigned to represent the average intensity of positive cells (0: no staining; 1: weak, 2: intermediate; 3: strong). The proportion score and intensity score were added to get a final score (from 0 to 8).
Three antibodies for ERβ were used for Western blot and immunohistochemistry: ab H150 (polyclonal, Santa Cruz), ab 1531 (monoclonal, Santa Cruz), ab 3576 (polyclonal, Abcam). Isoform specific antibodies for ERβ1, ERβ2 and ERβ5 were purchased from AbD Serotec (Oxford, UK). Other antibodies used are: actin (monoclonal, Santa Cruz), flag (monoclonal, Sigma), PTEN (monoclonal, Santa Cruz). Antibodies for the PI3K/AKT and MAPK/ERK pathways were purchased from Cell Signaling (Boston, USA). Human glioblastoma multiforme cell lines U87 and A172 were purchased from ATCC. Human primary astrocytes were kind gifts from Dr. Anuja Ghorpade in the department of Cell Biology & Anatomy in University of North Texas and were prepared as describe previously (Fields et al., 2011).
RNA was extracted using Trizol (Invitrogen) following manufacturer’s protocol and quantitated using Nanodrop (Thermo Scientific). One μg of RNA was used for cDNA synthesis. Isoform specific primers for ERβ were designed as described in the previous publication by Moore et al. (1998) for PCR. Shared forward primer: 5′-AGT ATG TAC CCT CTG GTC ACA GC G-3′; Reverse primers: ERβ1: 5′-CCA AAT GAG GGA CCA CAC AGC AG-3′, ERβ2: 5′-GGA TTA CAA TGA TCC CAG AGG GAA ATT G-3′, ERβ3: 5′-GCA GTC AAG GTG TCG ACA AAG GCT GC-3′, ERβ4: 5′-GGA TTA CAA TGA TCC CAG AGG GAA ATT G-3′, ERβ5: 5′-CTT TAG GCC ACC GAG TTG ATT AGA G-3′; actin forward: 5′-CCA ACA CAG TGC TGT CTG G-3′, actin reverse: 5′-TGC TGA TCC ACA TCT GCT G-3′. PCR products were analyzed in 2% agarose gel. Another set of isoform specific primers by Leung et al. (2006) were also used for regular PCR and real-time PCR detection of ERβ isoforms: ERβ1 forward: 5′-GTC AGG CAT GCG AGT AAC AA-3′; ERβ1 reverse: 5′-GGG AGC CCT CTT TGC TTT TA-3′; ERβ2 forward: 5′-TCT CCT CCC AGC AGC AAT CC-3′; ERβ2 reverse: 5′-GGT CAC TGC TCC ATC GTT GC-3′; ERβ4 forward: 5′-GTG ACC GAT GCT TTG GTT TG-3′; ERβ4 reverse: 5′-ATC TTT CAT TGC CCA CAT GC-3′; ERβ5 forward: 5′-GAT GCT TTG GTT TGG GTG AT-3′; ERβ5 reverse: 5′-CCT CCG TGG AGC ACA TAA TC-3′; GAPDH forward: 5′-TCC CTG AGC TGA ACG GGA AG-3′; GAPDH reverse: 5′-GGA GGA GTG GGT GTC GCT GT-3′. Real-time PCR was carried out by a 7300 Real-Time PCR System (Applied Biosystems) using SYBR Green PCR Master Mix (Promega).
ERβ1 (530 aa) and ERβ5 (472 aa) cDNA sequences with a flag tag (GAT TAC AAG GAT GAC GAC GAT AAG) at N-terminal were inserted into pCDNA3.1 (Invitrogen) between cloning sites XhoI and HindIII. Plasmids were transfected using lipofectamine 2000 (Invitrogen) following the manufacturer’s protocol. Forty-eight hours after transfection, G418 was added to the medium (2 mg/ml) for selection of stable clones. Medium was changed every two days. G418 resistant clones were selected for culture in medium with 200 μg/ml G418. Expression of ERβ was validated using both Western blot and immunocytochemistry.
Plasmids expressing constitutively active form of HIF1α and HIF2α, designated as P1P2N HIF1α and P1P2N HIF2α, were kind gifts from Dr. Joseph A. Garcia at University of Texas Southwestern Medical Center at Dallas (Dioum et al., 2008). The plasmids were generated by alanine substitution of the conserved proline or asparagine residues in HIF1α and HIF2α and cloned into pIRES-hrGFP vector (Dioum et al., 2008). Forty eight hours after transfection, real-time PCR was carried out to evaluated mRNA levels of ERβ1, ERβ2 and ERβ5.
HEK 293 stable clones were seeded in the Seahorse XF-24 plates. After 24 h, cells were changed to unbuffered DMEM (DMEM base medium supplemented with 25 mM glucose, 10 mM sodium pyruvate, 31 mM NaCl, 2 mM Glutamine, pH 7.4) and incubated at 37 °C in a non-CO2 incubator for 1 h. Eight baseline measurements of oxygen consumption rate (OCR) were collected. After seahorse analysis, cells in the wells were lysed using protein lysis buffer and protein concentrations were measured using protein assay (Thermo Scientific). Oxygen consumption of each well was normalized to total protein amount (pmol/min/μg).
U87 cells were cultured in phenol-red free medium (with 10% charcoal stripped FBS) at 25,000 cells per well in 12-well plates. At indicated days after seeding, cells were trypsinized and counted using a hemocytometer. Four wells were assigned to each group. Cell counting was conducted by a researcher who was blinded to the treatment/group assignment. Each growth curve was validated by duplication or triplication.
U87 cells were seeded in 6-well plate and cultured overnight. Medium was replaced with DMEM without FBS and culture for 6 h for cell cycle synchronization. Then 10% FBS was added to the medium. After 12 h and 24 h, cells were collected for propidium iodide (PI) staining. After staining, cells were analyzed using a BD LSR II Flow Cytometer.
Cells were lysed in RIPA buffer and protein concentration was measured using protein assay. Protein lysates were diluted to 1 μg/μl and mixed with non-reducing protein loading buffer. Gelatin zymography was done as described previously (Li et al., 2011). Briefly, 15 μl of sample was loaded to 10% polyacrylamide gel containing 0.1% gelatin. After electrophoresis, the gel was incubated at 37 °C in incubation buffer (2.5% Triton X-100, 5 mM CaCl2, 1 μM ZnCl2) for overnight with gentle agitation. The gel was stained using Coomassie Blue solution (0.25% Coomassie Blue, 45% MeOH, and 10% acetic acid) for 2 h and then destained with destain solution (30% MeOH and 10% acetic acid) until the bands became clear. Image was taken using UVP imaging system.
Values were expressed as mean±standard error of mean (SEM). Multiple comparisons were performed by one-way ANOVA. When a significant difference was detected by ANOVA, a post hoc Tukey’s test was performed to identify a specific difference between groups. Two-way ANOVA was performed to analyze growth curve data (days and clone/treatment). Between two groups, student’s t-tests were used to acquire a p value. A p value of *, p<0.05; or **, p<0.01 was used to indicate statistical significance.
We thank Dr. Joseph A. Garcia at University of Texas South-western Medical Center at Dallas for providing plasmids P1P2N HIF1α and P1P2N HIF2α.
Funding This work was partly supported by National Institutes of Health grants R01NS054687 (SY), R01NS054651 (SY).
Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.brainres.2013.02.004.