Background
Glioblastoma is the most common and lethal primary malignant brain tumor which remains a challenging disease to treat. The current standard-of-care for patients with newly diagnosed glioblastoma consists of maximal surgical resection, radiotherapy (RT), and concomitant and adjuvant chemotherapy with temozolomide (TMZ). Despite aggressive treatment, all patients eventually suffer from tumor progression because their tumors become resistant to maintenance TMZ, and the median survival among all patients is only 12–15 months from diagnosis. Although TMZ is the principal first-line chemotherapeutic agent used for the treatment of glioblastoma, it does not significantly prolong the overall survival of patients without methylation of the MGMT promoter [
1‐
3]. The anti-tumor activity of TMZ depends on its ability to methylate DNA at the O-6 positions of guanine residues, which will produce methylguanine adducts, triggering a continuous cycle of DNA base mismatch repair, which leads to double-strand breaks and base mispairing, ultimately inducing cell apoptosis [
4,
5]. MGMT is a cellular DNA repair protein that neutralizes the cytotoxic effects of TMZ by directly transferring methyl groups from the O-6-position of guanine to a cysteine residue [
6]. Therefore, glioblastoma tumors expressing MGMT have been implicated as a major intrinsic mechanism of resistance to TMZ, although a different mechanism independent of MGMT has been reported [
7]. Methylation of MGMT promoter has become an important prognostic and predictive factor for TMZ treatment of newly diagnosed GBM, and high MGMT protein expression in patient tumors is associated with TMZ resistance in patients [
8,
9]. Thus, treatment strategies to overcome MGMT-dependent or independent chemoresistance are urgently needed.
It has been hypothesized that glioblastoma stem cells (GSC) are responsible for post-treatment tumor recurrence because they are drug-resistant cells that can survive treatment and regenerate tumors [
10‐
13]. A recent study using a genetically engineered mouse model of glioma has provided direct evidence and demonstrated that a quiescent subset of endogenous stem-like glioma cells is located at the apex of a cellular hierarchy in tumor maintenance, and is responsible for tumor recurrence after TMZ therapy fails [
14]. This study thus supports the view that TMZ can only deplete the proliferative differentiated tumor population, but not the quiescent GSC. It is generally accepted that GSC are a small subset of slow-cycling stem-like glioblastoma tumor cells within a tumor tissue, that are capable of clonally self-renewing and growing as tumor spheres and migrating radially outward in culture, and reconstituting a tumor in mouse brain that recapitulates the histopathological features of the patient tumor from which the GSC were derived [
15‐
17]. A previous study indicated that MGMT-negative CSC line can be depleted with 50 μM TMZ treatment in culture whereas GSC line expressing MGMT transcripts results in a 10-fold increase of TMZ-resistance (500 μM) [
18]. Similarly, we also found that GSC clones resistant to radiochemotherapy expressed upregulated MGMT when compared to that of autologous sensitive GSC clones [
13]. Current adjuvant TMZ treatment is given as 150–200 mg/m
2 on days 1 to 5 of a 28-day cycle, which results in concentrations of between 15–35 μM in glioma tumor tissue [
19]. Therefore, identifying a new strategy to sensitize the MGMT-expressing GSC to clinically achievable dose of TMZ in brain will have important implications for the management of glioblastoma patients with unmethylated MGMT promoter.
In this study, we extended our previous work and used MGMT-expressing GSC that survived 500 μM TMZ treatment (GSC-500 μM TMZ) to explore the potential intrinsic factors that may be linked to triggering TMZ resistance. By comparing gene-expression profiles between GSC-500 μM TMZ and parental GSC (GSC-parental), we explored a series of genes that characterized intracellular stress responses and self-defense mechanisms against high-dose TMZ in GSC-500 μM TMZ. Moreover, BMP7 was identified as a top down-regulated gene in GSC-500 μM TMZ. We thus evaluated the treatment efficacy of recombinant BMP7 on GSC and tested the synergistic effect of BMP7 and low-dose TMZ on the treatment of MGMT-expressing GSC both in vitro and in vivo. We further investigated the potential MGMT-independent mechanisms contributing to the BMP7-mediated sensitization of MGMT-expressing GSC to TMZ.
Discussion
In this study, we used molecular profiles of MGMT-expressing GSC that survived high-dose TMZ treatment, to probe defense signatures which could be potential treatment targets for sensitizing GSC to the clinically relevant dose of TMZ. Using this non-biased strategy, we have identified informative gene profiles that are likely to contribute to resistance to high-dose TMZ treatment. These protective stress response profiles are similar to our previous finding in GSC clones that survived radiochemotherapy (RT + TMZ) [
13], which expressed molecular and functional characteristics resembling the anti-aging/anti-stress effects of caloric/glucose restriction (GR), by which both insulin-like growth factor 1 (IGF1) and insulin/Akt signaling were reduced [
32‐
34]. The transcription profiles suggested that the stress/drug resistance of GSC-500 μM TMZ is associated with cellular quiescence, EMT/invasiveness, suppressed growth and differentiation, and impaired insulin/Akt signaling. We unintentionally found BMP7 to be a top down-regulated gene in GSC-500 μM TMZ, and thus hypothesized that reduced BMP7 expression/signaling helped them maintain their dedifferentiated state, which prevents them from premature senescence, and renders them more resistant to standard treatment, since it only targets more differentiated/aging cells [
35]. Indeed, treatment with BMP7 alone allows for delaying tumor development/progression without TMZ compared to untreatment or treatment with TMZ alone, suggesting that the anti-tumor activity of BMP7 is independent of MGMT status in GSC. This notion is further supported by the finding that BMP7 treatment does not reverse the unmethylation status of GSC. Our data support the view that induction of cell senescence/aging and loss of EMT/migration/invasion properties by BMP7 treatment are likely to contribute to the reduction of tumorigenesis and progression, leading to prolonged animal survival. Moreover, BMP7 treatment down-regulates the expression of MGMT and ATP-binding cassette drug efflux transporters in GSC may provide an additional support mechanism to synergize low-dose TMZ in treatment of GSC. Therefore, BMP7 treatment induces cooperative mechanisms which allow for sensitization of GSC to low-dose TMZ treatment, and extends animal survival. Similar to our finding, a recent study showed that MGMT methylation status does not predict TMZ response in GSC model, and both methylated and unmethylated MGMT bands can be amplified in TMZ-resistant GSC lines [
36]. Another study found that some GBM lines resistant to TMZ treatment do not express MGMT protein, but rather exhibit a down-regulation of DNA mismatch repair protein or reduced methylation of LINE-1 repetitive elements (enhances transposon activity) [
37]. Therefore, we believe that down-regulation of MGMT expression by BMP7 is not the sole mechanism allowing for GSC responding to low-dose TMZ treatment.
BMP7 is a member of the transforming growth factor-β (TGF-β) superfamily of growth factors. The binding of BMP7 to its receptors, BMP type 1 and type 2 receptors (BMPR1/2), induces the phosphorylation of intracellular SMAD1/5/8, which can block the nuclear translocation of phosphorylated SMADs 2/3 induced by TGF-β, which in turn results in suppressed TGF-β signaling. BMP7 plays a pivotal role in the osteoblast differentiation/bone formation, kidney development, and promoting brown adipocyte differentiation [
38‐
40]. Recently, BMP7 has been implicated in regulation of cancer pathogenesis and metastasis, possibly due to its ability to counteract TGF-β-induced, SMAD3-dependent EMT [
41,
42]. Reduced levels of BMP7 in primary breast and lung cancer tissues are significantly associated with the formation of clinically overt bone metastases for breast cancer patients and lymph node metastasis for lung cancer patients [
43,
44]. Down-regulated BMP7 expression was also determined in primary human prostate cancer tissue when compared with normal prostate luminal epithelium [
45]. The animal studies further demonstrated that BMP7 treatment significantly inhibited internal bone growth of breast cancer cells and prostate cancer bone metastases, suggesting that reduced BMP7 signaling in tumor cells may enhance tumorigenesis and EMT with the development of metastatic properties via enhanced capacity for cell migration and invasion [
43,
45]. Our previous [
13] and current data support the view that EMT linked with dedifferentiation/stemness are possibly underlying treatment resistance of GSC, which can be overridden by augmenting BMP7 signaling. The removal of factors associated with stemness or anti-stress properties in GSC by BMP7 treatment is evident by demonstrating the significantly downregulation of mRNA levels of CD133, MGMT, and efflux transporters with concomitant induction of cell senescence and susceptibility to low-dose TMZ treatment. A previous study has shown that induction of EMT in mammary epithelial cells by exposure to TGF-β1 or overexpression of Snail or Twist (EMT-inducing transcription factors), can generate cells with stem cell properties [
46]. Another study also found that ZEB1, an EMT activator, represses expression of stemness-inhibiting microRNA [
47]. Likewise, we previously found that primary glioblastoma possess molecular properties of mesenchymal stem cells (MSC) and express cellular and molecular markers that have been implicated in EMT/myofibroblastic phenotype [
48]. In particular, we found primary glioblastoma tumor cells express CD105 (endoglin), which is an MSC surface marker and a component of the TGF-β receptor complex that binds to TGF-β1 and TGF-β3, and modulates TGF-β signaling [
48], thus suggesting a direct link between TGF-β-induced EMT and the gain of stem cell properties in glioblastoma and GSC. TGF-β1 promotes hematopoietic stem cell quiescence by downregulating Akt activity and upregulating FOXO3 activity [
49]. Likewise, we also found that GSC clones sensitive to radiochemotherapy (RT + TMZ) exhibited activated Akt activity with increased glucose usage, whereas resistant clones expressed upregulated CD133, SOX2 and MGMT, with reduced Akt activity and increased AMPK-SIRT1-FOXO Axis and favored the fatty-acid oxidation pathway for their energy source [
13]. In this study, we also found that treatment with BMP7, an antagonist of TGF-β system, downregulated EMT and stemness transcription program accompanied by upregulation of genes associated with insulin signaling/AKT and cell senescence, suggesting that BMP7 modulates stemness and EMT may be involved in a metabolic switch, which led to increased drug sensitivity. Similarly, a recent study showed that BMP2 can sensitize glioblastoma-stem-like cells to TMZ (500 μM) by downregulating both hypoxia-inducible factor-1α (HIF-1α) and MGMT [
50]. The study also demonstrated a direct binding of HIFs to the MGMT promoter under hypoxia, and the treatment with BMP2 can abrogate HIF-1a binding to MGMT promoter [
50]. The hypoxia/HIF-dependent promotion of the stemness/cellular quiescence and EMT in cancer progression has been well addressed and reviewed [
51‐
53]. Correspondingly, it was reported that ABC efflux transporters contain several binding sites for EMT-inducing transcription factors, and overexpression of Twist, Snail, and FOXC2 can increase promoter activity of ABC efflux transporters by binds directly to the E-box elements of ABC efflux transporters [
54]. These findings therefore support the notion that BMP7-mediated downregulation of CD133, MGMT and efflux transporters accompanied with increase sensitivity to TMZ may be achieved via the removal of factors that promote EMT and stemness properties in GSC.
We previously reported that GSC clones surviving radiochemotherapy (RT + TMZ), or purified CD133
+ GSC, expressed upregulated angiogenesis- and EMT-associated genes [
13,
17]. In this study, we further found that BMP7 treatment markedly upregulates genes associated with cell differentiation and senescence while downregulates genes associated with stemness properties in GSC, including CD133 expression, suggesting that the expression of CD133 may be an indication of both stemness and drug resistance in GSC [
22‐
24]. Recent reports indicated that CD133 is a marker of bioenergetic stress in hypoxic human glioma [
55] and that activation of hypoxia/HIF-1alpha enhanced the self-renewal activity of CD133
+ GSC and inhibited the induction of CSC differentiation [
56]. Moreover, the expression of CD133 facilitates EMT [
57], whereas CD133 silencing inhibits stemness properties and enhances chemoradiosensitivity of tumor stem cells [
58]. In this study, we identified several reported EMT inducers as molecular signatures of GSC-500 μM TMZ, including MCAM, FN1, and MALAT1 [
59‐
61]. Correspondingly, NOTCH3, which both gates neural stem cell activation [
62] and contributes to TGF-β-induced EMT [
63], was detected as a downregulated gene by BMP7 treatment. Meanwhile, a NOTCH signaling repressor, ATXN1 [
64], was identified as an upregulated gene by BMP7 treatment. Thus, maintaining the low activity of BMP7 signaling in GSC may be required for sustaining EMT, stemness, and multidrug-resistant phenotype. Our finding is in agreement with that of others showing that BMP7 release from endogenous neural precursor cells can induce tumor stem cell differentiation and reduce the ability for tumor initiation, therefore, providing a protective action in animals [
65]. Likewise, treatment with a BMP7 variant suppressed the tumorigenicity of stem-like glioblastoma cells and reduced angiogenesis and brain invasion [
66]. The optical imaging study further provided direct evidence of BMP7-induced cell cycle arrest in glioma model [
67]. Notably, it was reported that BMP7-induced senescence of prostate cancer stem-like cells is reversible; withdrawal of BMP7 treatment restarted growth of these cells in bone [
68]. Therefore, combination with chemotherapy would provide extended survival time by eliminating the cells, preventing them from regrowing back. Genetic approaches for elucidation of mechanisms by which BMP7 downregulates MGMT, ABC efflux transporters, stemness, and EMT would further facilitate our understanding the process and ability to identify new treatment targets and strategies for preventing treatment resistance and tumor recurrence.
Methods
Cell cultures
Glioblastoma stem cell (GSC) cultures used in this study were established from glioblastoma tumor tissues derived from patients who underwent surgery at Ronald Reagan UCLA Medical Center. All samples collected were under patients’ written consent, and were approved by the UCLA Institutional Review Board. The tumors were enzyme-digested and washed, followed by red blood cell (RBC) lysis of the pellet. The primary cells were cultured in a serum-free stem cell culture medium consisting of DMEM/Ham’s F-12 (Mediatech, Manassas, VA), 20 ng/ml human recombinant epidermal growth factor (EGF, Sigma-Aldrich, St. Louis, MO), 20 ng/ml basic fibroblast growth factor (FGF, Chemicon, Billerica, MA), 10 ng/ml leukemia inhibitory factor (LIF, Chemicon), and B27 without vitamin A (Invitrogen, Carlsbad, CA). The tumor spheres were dissociated and replated at clonal density and continually passaged until the clonogenic cells were stably maintained. As previously described [
13,
17], the D431 GSC culture line was classified as mesenchymal subtype whereas S496 and E445 culture lines were classified as proneural (PN) subtype. All three tumorigenic GSC cultures contained CD133
+ cells (39.5 %, 9.6 % and 1.5 % respectively), exhibited wild-type IDH1/IDH2 and unmethylated MGMT promoter and expressed MGMT transcripts (10, 14, Additional file
1: Figure S3).
Isolation of clonogenic GSC resistant to high-dose TMZ
GSC cultures were seeded at clonal density overnight, and treated with 500 μM TMZ the next day. Fresh media was replaced every three days. TMZ treatment was repeated on day 7 after the first treatment to ensure that clonogenic survivors are truly resistant to 500 μM TMZ. Clonogenic cells that have survived were isolated, expanded, and designated as GSC-500TMZ-C1 (treatment cycle 1). Established GSC-500TMZ-C1 cultures were then reseeded and re-treated with 500 μM TMZ twice (day 1 and day 7), and the stable clonogenic cells continued to grow were harvested and designated as GSC-500TMZ-C2. In order to maintain resistant clones in the long-term cultures, cells were re-exposed to 500 μM TMZ treatment after thawing or prior to using in experiments.
Cell proliferation
The effects of TMZ or siRNA treatment on GSC growth were determined by a 3-(4, 5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS/PMS) colorimetric assay according to the manufacturer’s instructions (Promega).
Cells were seeded in 96-well tissue culture plates at a density of 6,000–10,000 cells per well per 100 μL stem cell media in triplicate in the presence or absence of indicated treatment. Cells were incubated for 48–72 h, and absorbance was measured at 490 nm after a 4-hour incubation with MTS/PMS reagents.
Cell apoptosis
Cell apoptosis in BMP7- and TMZ-treated GSC was determined by Apo-ONE® Homogeneous Caspase-3/7 Assay kit (Promega), according to the manufacturer's protocol. One hundred microliters Homogeneous Caspase-3/7 Reagent was added to each well followed by 2 min mixing on a plate shaker. Plates were incubated at room temperature for 2 h. Fluorescence intensity was measured with a fluorescence microplate reader (Synergy HT, BioTek) at the excitation and emission wavelengths of 485 nm and 528 nm respectively. Multiple readings were taken at one-hr intervals. Data are expressed as the relative fluorescence units (RFU).
Cell-cycle distribution
2–5 × 105 dissociated cells were washed twice with cold PBS. Cell pellets were resuspended in 1 ml propidium iodide hypotonic DNA staining buffer (50 mg/ml propidium iodide, 0.1 % Triton X-100, and 0.1 % sodium citrate in PBS) and mix well. Samples were kept in 4 °C away from light for a maximum of 1 h before acquisition on the flow cytometer for cell cycle analysis (FACScan, Becton Dickinson), which used DNA content as a measure of progression in cell cycle and as means of detecting apoptotic cells. Apoptotic cells with degraded DNA were detected as a hypodiploid or "sub-G1" peak in a DNA histogram.
Microarray procedures, data analysis and gene annotation
Molecular profiling and analysis were performed as described [
17]. Briefly, cDNA was generated and converted to cRNA probes using standard Affymetrix protocols and hybridized to Affymetrix GeneChip U133 Plus 2.0 Array. The chips were scanned using the GeneArray scanner (Affymetrix). The CEL files generated by the Affymetrix Microarray Suite version 5.0 were converted into DCP files using the DNA-Chip Analyzer (dChip/2008;
http://www.hsph.harvard.edu/cli/complab/dchip/). The DCP files were globally normalized, and gene expression values were generated using the dChip implementation of perfect-match minus mismatch model-based expression index. All group comparisons were performed in dChip. Functional annotation of individual genes was obtained from NCBI/Entrez Gene (
http://www.ncbi.nlm.nih.gov/sites/entrez), the published literature in PubMed Central (NCBI/PubMed), GeneCards (
http://www.genecards.org/), and Protein knowledge base (UniProtKB) (
http://beta.uniprot.org/). All microarray CEL files analyzed in this study are accessible from the Gene Expression Omnibus (GEO) (Series Accession number: GSE68071).
Semi-quantitative reverse transcriptase polymerase chain reaction (sqRT-PCR) analysis
Cells were subjected to total RNA extraction using RNeasy kit (Qiagen, Valencia, CA). Two micrograms of total RNA from each sample were reverse transcribed to cDNA using a TaqMan RT Reagent Kit (Applied Biosystems). Thirty cycles of PCR amplification was performed on an Eppendorf gradient thermocycler, using 5 μL cDNA, SYBR Green PCR Core Reagents (Applied Biosystems) and gene-specific primers (Invitrogen). The signal intensity of each specific gene was quantified using Image Lab™ Software (Gel Doc™ EZ System, Bio-Rad). The band intensity of each sample was normalized to the corresponding β-actin band intensity in order to obtain the relative level of gene expression. The primer sequences and expected sizes of amplified PCR products are described in Additional file
1: Table S1.
Western blot analysis
30 μg protein from each sample were separated on 4–20 % gradient SDS-PAGE (Bio-Rad) and transferred onto a PVDF membrane. The blots were incubated with phospho-Smad1 (Ser463/465)/ Smad5 (Ser463/465)/Smad9 (Smad8) (Ser465/467) antibody (Cell Signaling) for overnight at 4 °C. Anti-β-actin (Cell Signaling) was used as internal control. The blots were washed and incubated with horseradish peroxidase-conjugated anti-rabbit IgG for 1 h. After washing, blots were incubated with Pierce Supersignal ECL substrate, and exposed to X-ray films.
siRNA transfection
A reverse transfection protocol was done to deliver non-silencing negative control siRNA (scrambled siRNA) (Ambion), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or gene-specific siRNA (Ambion), into GSCs as previously described [
13]. The transfection efficiency and cellular toxicity due to transfection were monitored using KDalert™ GAPDH Assay Kit (Invitrogen). Briefly, a transfection complex was prepared by diluting siRNA in 10 μl OPTI-MEMI (Invitrogen) then mixing with 10 μL OPTI-MEMI containing 0.3 uL Lipofectamine RNAi-MAX transfection reagent (Invitrogen). The siRNA transfectant was then added into each well in a 96-well plate followed by seeding 6000–9000 cells in 100 μL media to give a final siRNA concentration of 30 nM in each well. Targeted gene silencing was determined 72 h after transfection by sqRT-PCR, using a Power SYBRH Green Cells-to-CTTM Kit (Ambion).
Methylation-specific PCR (MSP) analysis
Genomic DNA was subjected to bisulfite treatment using the EZ DNA Methylation-Gold™ Kit (Zymo Research) following the manufacturer's instructions. The promoter MGMT-MSP was performed using a two-step nested approach to amplify both methylated and unmethylated MGMT separately as described previously [
71]. Total human genomic DNA methylated by bacterial DNA methyltransferase and whole genome amplified DNA were used as positive and negative controls for methylated alleles of MGMT, respectively. The PCR products were resolved on 4 % low melting point agarose gels.
To test the tumorigenic capacity of GSC, 105 viable cells in a volume of 3 μl culture media were engrafted intracranially into anesthetized NOD (CB17-Prkdcscid/J) mice (5 weeks old; 15–16 g). Mice were then maintained until neurological signs were observed, at which point they were sacrificed. The brains were removed, fixed in 4 % formalin, paraffin-embedded, and sectioned. Histopathologic analyses were done on paraffin slides stained with hematoxylin-eosin (H-E) staining per standard technique. To test whether BMP7 can sensitize TMZ for treatment of GSC in vivo, mice were randomly divided into four groups (5 mice per group) and each group was received the following treatments: group 1, 0.01 % DMSO; group 2, TMZ; group3, BMP7; group 4, BMP7 + TMZ. GSC cells (105/3 μl/animal) were intracranially injected followed by local delivery of vehicle (0.01 % DMSO in 100 μl H2O) or TMZ (100 μl of 35 μM TMZ) for group 1 and group 2 mice or BMP7 (10 ng in 100 μl water) for group 3 and group 4 mice using an osmotic minipump (Alzet, USA, model 1007D). The sterile pump was implanted subcutaneously (s.c.) onto the back of animals, and the drug solutions were delivered through the cannula, which was placed through a small skull burr hole onto the pial surface where tumor cells were implanted. On day 7 after local treatment, the mice of group 2 were continuously treated with TMZ (66 mg/kg) via oral gavage daily for 5 days, the mice of group 3 received intraperitoneal injection of BMP7 (2 μg/day) for 5 days and mice group 4 received both BMP7 (morning) and TMZ (afternoon) for 5 days. Animals were maintained until neurological signs were observed. Immediately after sacrifice, the brains were removed and subjected to histopathological analysis. All animal experiments in this study were under a protocol approved by the UCLA Institutional Animal Research Committee.
Statistical analysis
Each experiment was set up in triplicate and repeated at least twice. Data were expressed as means ± SD and analyzed using 1-way ANOVA tests, depending on homogeneity of variances. Cumulative survival probabilities were calculated using the Kaplan-Meier method. The log-rank test was used to compare survival across groups. All p-values were 2-sided, and those lower than 0.05 were considered significant. SPSS v19.0 for Windows software was used for all statistical analysis.
Competing interests
The authors have declared that no competing interests exist.
Authors’ contributions
JLT, SY, KY, WM, CLT conceived and designed the experiments. JLT, SY, JCM, KY, YZ conducted and managed animal experiments. JLT, JCM, IH, YB, AS carried out the cellular and molecular studies. KY and WHY performed pathology analysis. LML, SFN, TFC, WHY, AL contributed patient tumor samples and provided tumor information, analysis tools, and MGMT methylation status. JLT and CLT wrote the manuscript and performed statistical analysis. All authors read and approved the final manuscript.