Abstract
Glioblastoma multiforme (GBM) displays high resistance to radiation and chemotherapy, due to the presence of a fraction of GBM stem-like cells (GSLCs), which are thus representing the target for GBM elimination. Since mesenchymal stem cells (MSCs) display high tumor tropism, we examined possible antitumor effects of the secreted factors from human MSCs on four GSLC lines (NCH421k, NCH644, NIB26, and NIB50). We found that conditioned media from bone marrow and umbilical cord-derived MSCs (MSC-CM) mediated cell cycle arrest of GSLCs by downregulating cyclin D1. PCR arrays revealed significantly deregulated expression of 13 genes associated with senescence in NCH421k cells exposed to MSC-CM. Among these, ATM, CD44, COL1A1, MORC3, NOX4, CDKN1A, IGFBP5, and SERPINE1 genes were upregulated, whereas IGFBP3, CDKN2A, CITED2, FN1, and PRKCD genes were found to be downregulated. Pathway analyses in GO and KEGG revealed their association with p53 signaling, which can trigger senescence via cell cycle inhibitors p21 or p16. For both, upregulated expression was proven in all four GSLC lines exhibiting senescence after MSC-CM exposure. Moreover, MSC paracrine signals were shown to increase the sensitivity of NCH421k and NCH644 cells toward temozolomide, possibly by altering them toward more differentiated cell types, as evidenced by vimentin and GFAP upregulation, and Sox-2 and Notch-1 downregulation. Our findings support the notion that MSCs posses an intrinsic ability to inhibit cell cycle and induce senescence and differentiation of GSLCs.
Keywords
Introduction
Glioblastoma multiforme (GBM) is the most malignant type of brain tumor that is still incurable, with the overall survival of GBM patients being less than 15 months (35). This is due to highly infiltrative growth of GBM and its resistance to chemo/radiotherapies, preventing complete elimination of tumor cells, despite improvements made in surgical techniques and therapeutic protocols (7, 45). The highly tumorigenic subpopulation of glioblastoma stemlike cells (GSLCs) present in the tumor and defined as a subpopulation of cells with stem-like properties, such as self-renewal, asymmetric division, and differentiation potential, essential for tumor development and propagation in vivo (8, 45) are presumably causative for GBM recurrence. Their variable abundance within the tumor mass has been confirmed in several studies (6, 42, 52). Their stemness and increased resistance to radiation and chemotherapy (4, 16, 28) are suggesting new therapeutic approaches focusing on GSLC targeting to improve the survival of GBM patients (35).
Cell therapy is being revisited in view of using normal tissue stem cells, that is, human mesenchymal stem cells (MSCs) for cancer treatment. MSCs are multipotent stem cells most often isolated from birth-associated [i.e., umbilical cord (UC-MSC), cord blood (UCB-MSC)] and adult tissues, such as bone marrow (BM-MSC) and adipose tissue (AT-MSC) (31). They are supposed to have a wide therapeutic potential because of their immunomodulatory ability, wound- and neoplasma-directed homing, and tissue repair ability, relying on their differentiation potential (21, 24). Already naive UCB-MSCs were shown to suppress growth of various cancers, including GBM, as a 36% size reduction of C6 xenografts in mice was observed after UCB-MSC intratumoral injection (19). Likewise, a systemic AT-MSC coadministration with 8MGA glioma cells was shown to decrease de novo tumor development in mice by 63%, compared to control mice injected with 8MGA cells only (27). Injected AT-MSCs were shown to colocalize with CD133+ cells in GBM xenografts (2). Also in rat glioma models, the injection of UCB-MSCs and BM-MSCs resulted in tumor volume reduction (23), tumor growth inhibition (13), and increased survival of animals (33). Conditioned media (CM) derived from UC-MSCs and AT-MSCs were shown to inhibit the growth of U251 glioma cells (50), whereas BM-MSCs and their CM were shown to inhibit proliferation of U87MG, U251, and U373 glioma cells (31) and induce U87 cell senescence (32). In contrast, the UCB-MSCs, UC-MSCs, and AT-MSCs were shown to induce apoptosis in glioma cells (13, 20, 50). All of the above is implying the potential of MSCs for the development of novel anticancer cell-based therapies.
As the intrinsic resistance of GSLCs to therapy presents the main obstacle in GBM treatment, the repressive potential of MSCs on GSLCs (i.e., on their stemness potential and viability) should be defined. To clarify the MSC effect on GSLCs, we utilized the established GSLC lines and primary culture of GBM cells enriched for GSLCs and exposed them to BM-MSC- and UC-MSC-derived CM. We speculated that the paracrine activity of the MSCs, namely their secreted factors, would cause the alteration of the GSLC behavior. Indeed, we observed both MSC-CMs to cause cell cycle arrest of GSLCs in the G0/G1 phase and to induce their senescence. To our knowledge, we are also first to demonstrate that the exposure of GSLCs to MSC paracrine signals changes the gene expression of GSLCs, which is driving them toward a more differentiated phenotype. The MSC-CM also increased the sensitivity of GSLCs to chemotherapeutic temozolomide (TMZ), commonly used in GBM treatment.
Materials and Methods
Cell Cultures
GSLC lines NCH421k (male, age 66) and NCH644 (male, age 66) were isolated from GBM resection and checked for stemness potential, as described previously (8, 10).
NIB26 (female, age 75) and NIB50 (male, age 81) cells were isolated from fresh GBM samples obtained from the University Medical Centre of Ljubljana (study approved by the National Ethics Committee). Upon mechanical and enzymatic dissociation, the isolated cells were grown as spheroids (with a presumed enriched fraction of GSLCs) under the same conditions as NCH cells in neurobasal (NBE) medium (Gibco, Life Tech. Corp., Paisley, UK) supplemented with 2 mM l-glutamine (PAA Lab, Pasching, Austria), 100 U penicillin, 1,000 U streptomycin (PAA Lab), 1× B-27 (Gibco), 1 U/ml heparin (Sigma-Aldrich, Steinheim, Germany), 20 ng/ml EGF in 20 ng/ml bFGF (Gibco). Both derived cell lines were tested for the CD133 expression by flow cytometry.
The U87 cell line was grown as described previously (31).
Human MSCs
Human BM-MSCs were purchased from Lonza Bioscience (Walkersville, MD, USA) (BM-MSC2: male, age 19, 6F4393; BM-MSC3: female, 22, 7F3677) and cultured according to the manufacturer's recommendations. Briefly, MSCs were grown in Dulbecco's medium (DMEM; 5921; Sigma-Aldrich) with 10% fetal bovine serum (FBS; PAA Lab), supplemented with 100 U penicillin (PAA Lab), 1,000 U streptomycin (PAA Lab), 2 mM l-glutamine (PAA Lab), Na-pyruvate (Gibco), and nonessential amino acids (Sigma-Aldrich).
UC-MSCs were isolated from Wharton's jelly according to standard protocol (47) in the study approved by the National Ethics Committee, Doc. No. 134/01/11. Umbilical cords were collected at cesarean section (37-11 weeks) upon obtained informed consent, and isolated UC-MSCs were cultured the same as BM-MSCs. UC-MSCs were characterized for CD13+, CD29+, CD44+, CD73+, CD90+, CD105+, CD14-, CD34-, CD45-, HLA-DR- surface marker expression, and osteogenic, chondrogenic, and adipogenic differentiation as recommended (15). Two UC-MSC clones (UC-MSC-10, female and UC-MSC-13, male) with the highest proliferation potential and homogenous spindle-like morphology were used in further experiments.
Conditioned Media (CM) Collection
Two different clones of BM-MSCs and the two clones of UC-MSCs were plated into T75 flasks (Corning, Cambridge, MA, USA) at a density of 6,000 cells/cm2, and a day before the cells would have reached 90% confluence, the standard MSC cultivation medium, described above, was changed to 12.5 ml NBE medium (used in GSLC culture). After 24 h, this CM was collected and pooled from both BM-MSC lines or from both UC-MSC cell lines, centrifuged at 300 × g for 10 min at 4°C, and stored at −80deg;C. Prior to the experiment, the CM were mixed in a 1:1 ratio with fresh NBE medium and designated as BM-CM and UC-CM, respectively. Similarly, the NBE CM were collected after 24 h from cultured NCH421k and NCH644 spheroids and diluted at a 1:1 ratio with fresh NBE medium and when used designated as NBE-CM.
Surface Marker Detection by Flow Cytometry
UC-MSCs (106) were harvested, washed with 1× phosphate-buffered saline (PBS; Gibco), and incubated with antibodies against CD13 (BD#557454), CD29 (BD# 559883), CD44 (BD#555479), CD73 (BD#550257), CD90 (BD#555596), CD14 (BD#555397), CD34 (BD#555821), CD45 (BD#555482), HLA-DR (BD#340688) (all from BD Biosciences, San Jose, CA, USA), CD105 (MHCD1 0505; Molecular Probes, Eugene, OR, USA), and isotype controls [FITC-IgG1 (BD#555748), FITC-IgG2a (BD#55 5573), PE-IgG1 (BD#555749), PE-IgG2b (BD#555743), APC-IgG1 (BD#555751); all from BD Biosciences] as instructed by the manufacturer. All cells were additionally stained with propidium iodide (PI; BD Biosciences) to exclude dead cells, washed, and analyzed by flow cytometry using a BD FACSCalibur™ and the CellQuest Software (both BD Biosciences).
NIB26 and NIB50 cells were tested for the expression of CD133 protein using the antibodies recognizing AC133 or CD133/2 (AC141) epitope by the protocol described below. The staining results proved to overlap properly (8).
NCH421k and NCH644 (106) cells were harvested upon 72-h exposure to NBE, NBE-CM, UC-CM, and BM-CM, washed with 1× PBS, and incubated with 10 μl CD133/2-PE antibody (130-090-853, Miltenyi Biotec, San Diego, CA, USA) or 20 μl isotype control antibody (PE-IgG2b, BD Biosciences) in equal concentrations. All cells were PI (BD Biosciences) stained for live cell analysis (20 min at 4°C), washed, and analyzed by flow cytometry using a BD FACSCalibur™ and the CellQuest Software (both BD Biosciences).
Differentiation of UC-MSCs
Adipogenic Differentiation
The adipogenic differentiation medium was composed of DMEM supplemented with 10% FBS, 1 μM dexamethasone (Sigma-Aldrich), 0.5 mM 3-isobutyl-1-methyl-xanthine (IBMX; Sigma-Aldrich), 10 μg/ml insulin (Sigma-Aldrich), 100 μM indomethacin (Sigma-Aldrich). After 72 h, cells were medium changed to adipogenic maintenance medium: DMEM supplemented with 10% FBS and 10 μg/ml insulin for 24 h, and this sequence was repeated three times, followed by cultivation for 1 week in maintenance medium. After 21 days, cells were fixed in 4% paraformaldehyde for 10 min, and accumulated lipid-rich vacuoles were stained with 0.3% Oil red O (Sigma-Aldrich) for 30 min.
Osteogenic Differentiation
The osteogenic differentiation medium was prepared with DMEM containing 10% FBS, 1 μM dexamethasone (Sigma-Aldrich), 50 μg/ml l-ascorbic acid (Sigma-Aldrich), and 10 mM glycerophosphate (Sigma-Aldrich). After 21 days, cells were fixed in 4% paraformaldehyde for 10 min and stained with 1% Alizarin red S (Sigma-Aldrich) for 30 min to detect calcified extracellular matrix.
Chondrogenic Differentiation
Cells were exposed to chondrogenic differentiation medium composed of DMEM, 10% FBS, 10 ng/ml TGF-β3 (Sigma-Aldrich), and 50 μg/ml l-ascorbic acid for 21 days. To detect matrix deposition of sulfated glycosaminogycans (GAGs), cells were fixed in 4% paraformaldehyde for 10 min and stained with 1% Alcian blue 8-GX (Sigma-Aldrich) in 3% acetic acid (Sigma-Aldrich).
Cell Cycle Analysis
NCH421k cells were exposed to NBE, UC-CM, and BM-CM for 72 h, trypsinized [0.25% Trypsin-EDTA (Gibco)], washed with 1× PBS, and left in 75% ethanol overnight at 4°C. Upon washing with 1× PBS, cells were incubated in 0.5 ml PI/RNase staining buffer (BD Pharmingen, San Diego, CA, USA) for 15 min at RT. Cell cycle analysis was performed using the WinMDI and Cyclored on the BD FACSCalibur™ with the CellQuest Software.
Apoptosis Assay
NCH421k cells were exposed to NBE, UC-CM, and BM-CM for 72 h, harvested, and early/late apoptotic cells detected by staining with Annexin-V-FITC and PI (BD Pharmingen) according to the manufacturer's protocol. Stained cells were analyzed using BD FACSCalibur™ and the CellQuest Software. Staurosporine (STS; Sigma-Aldrich) treatment (2 μM, 6 h) served as a positive control for apoptosis.
Mitochondrial Membrane Potential (MMP) Detection
MMP change in NCH421k cells grown in NBE, UC-CM, and BM-CM for 72 h was measured with the Mitochondrial Permeability Transition Detection Kit JC-1 (ImmunoChemistry Technologies, Bloomington, MN, USA) according to the manufacturer's instructions. STS (2 μM, 6 h) and carbonylcyanide-chlorophenyl-hydrazone (CCCP; 25 μM, 30 min; ImmunoChemistry Technologies)-treated NCH421k cells were used as a positive control. JC-1 staining was analyzed using a Synergy™ MX Microplate Reader (Bio-Tec Instruments, Winooski, VT, USA). The change of MMP was assessed by the ratio of red versus green fluorescence readings.
Cell Senescence
NCH421k, NCH644, NIB26, and NIB50 spheroids were grown in NBE, UC-CM, and BM-CM for 72 h; GSLC spheroids were dissociated to single cells and fixed in 0.5% glutaraldehyde solution (Sigma-Aldrich) for 20 min at RT. The proportion of cells positive for senescence indicative of β-galactosidase (SA-β-Gal) activity was evaluated as described previously (14).
Quantitative Real-Time PCR (qRT-PCR) and PCR Arrays
Total RNA was isolated from three biological replicates of the U87 cell line and NCH421k, NCH644, NIB26, and NIB50 cells grown in NBE, UC-CM, and BM-CM for 72 h using Trizol™ reagent (Invitrogen Limited, Paisley, UK) following the manufacturer's instructions. Total RNA (1 μg) was used for cDNA generation using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA). Gene expression of cyclin-dependent kinase inhibitor 2A (CDKN2A; p16), cyclin-dependent kinase inhibitor 1A (CDKN1A; p21), cyclin D1 (CCND1), BCL-2-associated X protein (BAX), B-cell CLL/lymphoma 2 (BCL-2), prominin 1 (PROM1; CD133), SRY (sex-determining region Y)-box 2 (Sox-2), nestin (NES), Notch-1, tubulin β 3 class III (TUBB3), vimentin (VIM), and glial fibrillary acidic protein (GFAP) was quantified by real-time quantitative PCR on ABI 7900 HT Sequence Detection System (Applied Biosystems). Real-time PCR reactions comprised of cDNA added to TaqMan Universal PCR Master Mix and TaqMan Gene Expression Assays (all Applied Biosystems): CDKN2A, Hs00923894_m1; CDKN1A, Hs 00355782_m1; CCND1, Hs00765553_m1; BAX, Hs999 99001_m1; BCL-2, Hs0060823_m1; PROM1, Hs010092 50; Sox-2, Hs01053049_s1; NES, Hs00707120_s1; Notch-1, Hs01062014_m1; TUBB3, Hs00801390_s1; VIM, Hs001 85584_m1; GFAP, Hs00157674_m1; p53, Hs01034249; and glyceraldehyde 3-phosphate dehydrogenase (GADPH) TaqMan probe (Assay No. 4310884E) as an internal control. The analyses were performed with SDS v2.2 software (Applied Biosystems) and comparative Ct method (ΔΔCt algorithm).
By using generated cDNA from NCH421k exposed to control and both types of MSC-CM, the expression of 84 senescence-related genes was determined also following the Cellular Senescence RT-Prolifer PCR Array (Qiagen-SABiosciences, Hilden, Germany) protocol. On the Cellular Senescence RT-Prolifer PCR Array, the average of three different housekeeping genes, ACTB (actin β), GAPDH, and ribosomal protein L13a (RLP13A), was used in calculations as an internal control. Fold increase in mRNA levels was calculated with the ΔΔCt method by the provided online program available at http://pcrdataanalysis.sabiosciences.com/pcr/arrayanalysis.php. Results are expressed as means of three independent experiments done in duplicate.
Cytotoxicity Testing After Exposure to Temozolomide and 5-Fluoro-Uracil by MTS Cell Viability Assay
NCH421k and NCH644 cells were plated into 96-well plates (Corning) at a density of 104 cells/well and left to grow for 24 h before adding NBE, UC-CM, BM-CM, and TMZ (2.5-1000 μM) or 5-fluoro-uracil (5-FU) (0.01-150 μM) (both Sigma-Aldrich). After 48 h and 72 h of TMZ and 5-FU treatment, the CellTiter 96® AQueous MTS reagent [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium; Promega Corporation, Madison, WI, USA; Sigma-Aldrich], was added to each well to make 333 μg/ml final concentration. Three hours later, the absorbance at 490 nm (reference 690 nm) was measured with the Synergy™ MX Microplate Reader (Bio-Tec Instruments).
Statistical Analyses
All the above experiments were performed in duplicate and independently repeated three times. To test the effect of MSC-CM on GSLCs compared to control medium, the ANOVA test with post hoc Dunnett's test was performed using GraphPad Prism version 5.01 for Windows (GraphPad Software, La Jolla, CA, USA). A value of p < 0.05 was considered significant. Data are expressed as the mean ± standard deviation (SD).
Results
Umbilical Cord-Derived MSC Lines (UC-MSCs) and Primary GSLCs Were Successfully Generated
We established six MSC lines from umbilical cords (UC-MSCs) obtained from healthy donors. These UC-MSCs exhibited plastic adherence in culture and typical spindle-shaped morphology. We confirmed the presence of CD13, CD29, CD44, CD73, CD90, and CD105 and the absence of CD14, CD34, CD45, and HLA-DR surface antigens in all UC-MSC clones by flow cytometry and verified their osteo-, chondro-, and adipogenic differentiation potential (data available, not shown). This characterization is in accordance with suggested criteria for MSC phenotype confirmation (15); thus, we further used two UC-MSC clones together with two BM-MSC lines for UC-CM and BM-CM production.
Purified NCH421k and NCH644 GSLC lines were established previously by Campos et al. (10) and confirmed by DNA copy-number profiling and CGH array to show typical GBM chromosome aberrations (8, 46) and to highly express CD133 epitope 1 antigen detected by AC133 antibody (9). Flow cytometric analysis also confirmed expression of CD133 in NIB26 (79.9% CD133+ cells) and NIB50 (71.4% CD133+ cells) that were isolated from primary GBM tissue and grown in spheroids to enrich for their GSLCs fraction (Fig. 1A). Analysis of p53 status in NCH421k, NCH644, NIB26, and NIB50 cells was determined by qRT-PCR with U87 cell line (wild-type p53) serving as reference (54).
GSLCs isolated from primary GBM tissue express high levels of CD133. NIB26 and NIB50 cells were isolated from primary GBM tissue and grown in spheroids to enrich the GSLC fraction. Flow cytometry analysis of CD133 expression showed (A) 79.9% CD133+ cells in NIB26 and (B) 71.4% CD133+ cells in NIB50 cell culture. (C) Analysis of p53 gene expression in U87, NCH421k, NCH644, NIB26, and NIB50 cells by qRT-PCR.
MSC-CM Caused G0/G1 Phase Cell Cycle Arrest via Cyclin D1 Downregulation
Cell cycle analysis of NCH421k cells grown with NBE or either type of CM for 72 h was performed. It revealed an increased number of NCH421k cells halted in G0/G1 phase in the presence of UC-CM, compared to NBE (Fig. 2A). Additionally, the G0/G1 cell cycle arrest of NCH421k cells, grown in UC-CM and BM-CM, was confirmed by CCND1 gene expression analysis. This showed CCND1 gene expression to decrease 4.2-fold in UC-CM and 2.9-fold in BM-CM (Fig. 2B). Similarly, the CCND1 gene expression was found downregulated in NCH644, NIB26, and NIB50 cells exposed to CM. The above was associated with the indicative size reduction and loosened structure of NCH421k and NCH644 spheroids grown in both types of MSC-CM (Fig. 2C-E). The BM-CM was even shown to induce the adherent growth of NCH644 cells (Fig. 2F-H). Altogether, our data imply the capacity of MCS paracrine factors to cause the cell cycle arrest of GSLCs through decrease in the cyclin D1 expression level.
MSC-CM interferes with the cell cycle of GSLCs. (A) Cell cycle analysis of NCH421k cells performed by flow cytometry after 72-h treatment with both types of MSC-CM. (B) Cyclin D1 (CCND1) expression (qRT-PCR) was decreased in all GSLC lines after 72-h treatment with UC-CM and BM-CM. The mean ± SD of three independent experiments are provided. *p < 0.05, **p < 0.01, ***p < 0.001. Morphology of native cells: (C) Control NCH421k cells with compact spheroid morphology; (D) NCH421k spheroid morphology changed upon UC-CM treatment for 72 h; and (E) NCH421k spheroid morphology changed in the presence of BM-CM after 72 h. (F) Control NCH644 cells with compact spheroid morphology; (G) UC-CM treatment for 72 h changed NCH644 spheroid morphology to adherent type of growth; and (H) BM-CM presence after 72 h changed NCH644 spheroid morphology to adherent type of growth. Scale bar: 100 μm.
MSC-CM Affected Mitochondrial Membrane Potential but Did Not Induce Apoptosis in GSLCs
The induction of apoptosis in NCH421k cells grown in MSC-CM of both origins for 72 h was evaluated by MMP assay. A significant decrease in MMP was detected in NCH421k cells when exposed to UC-CM (23%) and BM-CM (31%) (Fig. 3A). Since MMP change is not solely indicative of apoptosis, BAX and BCL-2 gene expression was analyzed in the NCH421k, NCH644, NIB26, and NIB50 cells cultured in MSC-CM. The analysis of the BAX/BCL-2 gene ratio failed to demonstrate any significant change to occur in the CM-cultured GSLCs, compared to control (NBE) ones (Fig. 3B). To further exclude the onset of apoptosis, the annexin V staining was performed. This also failed to detect increased apoptosis in NCH421k cells cultured in MSC-CM versus control cells with staurosporin treatment used as positive control (Table 1). Altogether our results showed that by addition of MSC-CM containing MSC paracrine factors, the apoptosis in GSLCs cannot be induced.
MSC-CM does not induce apoptosis in GSLC lines. (A) Mitochondrial membrane potential (MMP) loss was ascertained by JC-1 staining after exposure of NCH421k cells to UC-CM and BM-CM for 72 h. (B) QRT-PCR analysis of Bax and Bcl-2 gene expression in NCH421k, NCH644, NIB26, and NIB50 cells after exposure to both types of MSC-CM for 72 h. The ratio between the Bax and Bcl-2 was unaffected. Data shown are mean ± SD of three independent experiments. ***p < 0.001. CCCP, carbonylcyanide-chlorophenyl-hydrazone; STS, staurosporin.
Annexin V/PI Staining of NCH421k Cells Exposed to Staurosporin (STS) and MSC-CM
The NCH421k cells were exposed to UC-CM and BM-CM, and the apoptotic cells were determined by flow cytometry as described in Materials and Methods. The exposure to STS, inducing intrinsic apoptotic pathway, was used as positive control. The data of three independent experiments are expressed as the mean ± standard deviation (SD) and *p < 0.05 was considered significant.
MSC-CM-Activated Senescence in GSLCs
As MSC-CM influenced the cell cycle of GSLCs, but did not cause apoptosis, the senescence in exposed GSLCs was investigated by using the indicative β-galactosidase staining. A significantly increased number of senescent cells was observed in NCH421k (17% and 33%), NCH644 (33% and 39%), NIB26 (20% and 25%), and NIB50 (30% and 31%) cells exposed to UC-CM and BM-CM, respectively, for 72 h, when compared to control NCH421k (Fig. 4A). In GSLCs exposed only to NBE, the percentage of senescent cells was 10.7% in NCH421k, 6.6% in NCH644, 3.3% in NIB26, and 8.2% in NIB50 cells.
MSC-CM activates senescence in all GSLC lines. (A) Senescence-associated β-galactosidase (SA-β-Gal) staining revealed activation of senescence in NCH421k, NCH644, NIB26, and NIB50 cells after exposure to UC-CM and BM-CM upon 72-h incubation. qRT-PCR analysis of (B) p21 expression and (C) p16 expression after 72-h treatment with both types of MSC-CM in all GSLCs. Data shown are mean ± SD of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001. (D) SA-β-Gal staining of dissociated and fixed NCH421k, NCH644, NIB26, and NIB50 cells after UC-CM and BM-CM presence for 72 h. Scale bar: 20 μm.
To get insight into the senescence signaling, NCH421k cells were subjected to both types of MSC-CM, and qRT-PCR analysis of 84 genes associated with senescence was conducted by the human Senescence RT2 Profiler PCR Array (Table 2). The PCR arrays revealed changed expression of 13 genes involved in senescence (ATM, CDKN1A, CDKN2A), cell adhesion (CD44, COL1A1), oxidative stress (PRKCD, NOX4), IGF-related pathway (IGFBP3, IGFBP5), p53/pRb signaling (MORC3, CITED2), and cytoskeleton formation (FN1, SERPINE1) (Table 2). Those genes were consistently up- and down-regulated in exposed NCH421k cells at least 2.1-fold. In addition, KEGG pathway analysis identified the involvement of five genes, SERPINE1, CDKN1A, IGFBP3, CDKN2A and ATM, in the p53 pathway (Table 3) and induced in senescence.
Differentially Expressed Genes in MSC-CM-Cultured NCH421k Cells
Listed are deregulated genes in NCH421k cells exposed to either UC-CM or BM-CM relative to NBE exposed NCH421k cells. Fold increase in mRNA levels are calculated as described in Materials and Methods. The mean ± SD of three independent experiments are provided.
p < 0.05. Fc, fold change.
Identification of Genes Involved in p53 Pathway With KEGG Pathway Analysis
Differentially expressed genes in NCH421k cells exposed to both types MSC-CM (Table 2) were subjected to ontology enrichment analysis in KEGG to detect their involvement in cellular pathways. Biomolecular identification number (Biomolecule IDs) corresponds to ENSEMBL gene description (http://www.ensembl.org); p value (<0.01); p FDR (false discovery rate, <0.01).
The PCR Array results were validated with qRT-PCR on independent mRNA samples of GSLCs, which confirmed increased expression of the CDKN1A gene in NCH421k cells grown for 72 h in UC-CM (1.4-fold) and BM-CM (6.7-fold). Likewise, the CDKN1A expression increased in NCH644 (3.2-fold) and NIB50 cells (5.0-fold) exposed to BM-CM (Fig. 4B). The expression of the CDKN2A gene also increased in NCH644 (3.4-fold), NIB26 (7.7fold), and NIB50 (1.9-fold) exposed to BM-CM (Fig. 4C), whereas it decreased 1.7-fold in NCH421k cells exposed to BM-CM (Fig. 4C). The above was consistent with the observed β-galactosidase staining (Fig. 4D), altogether implying the ability of the MSC paracrine factors to induce GSLC growth arrest and senescence.
MSC-CM Affected the Expression of GSLC Markers
To evaluate the impact of MSC-CM on the stemness of GSLCs, the expression of indicative GSLC marker CD133/prominin was examined by qRT-PCR (using gene expression assay recognizing all seven splice variants of the prominin transcript). An increased CD133 gene expression was observed in NCH421k cells (1.8-fold and 2.5-fold) and NIB26 cells (3.0-fold), whereas in NCH644 cells a decreased CD133 gene expression (2.3-fold and 3.9-fold) was noticed after 72 h of exposure to UC-CM and BM-CM, respectively (Fig. 5A). Flow cytometry analysis of CD133 glycosylated epitope (CD133/2) revealed a decrease in NCH644 cells exposed to UC-CM (59.5%) and to BM-CM (61.3%) for 72 h (Fig. 5B). In NCH421k cells exposed to both types of MSC-CM, flow cytometry analysis failed to demonstrate changes in CD133 antigen expression. Also, a prolonged culture of NCH421k cells in MSC-CM for 7 days had no impact on their CD133 antigen expression (Fig. 5B).
MSC-CM induces differentiation of GSLCs. (A) Analysis of CD133 expression by qRT-PCR after 72-h treatment with both types of MSC-CM was evaluated in all GSLCs. (B) Determination of CD133 protein expression by flow cytometry after exposure to UC-CM and BM-CM in NCH421k for 3 and 7 days and in NCH644 for 3 days. The qRT-PCR analysis after 72-h treatment with both types of 50% MSC-CM in NCH421k, NCH644, NIB26, and NIB50 cells for GSLCs markers expression (C) Sox-2, (D) Notch-1, (E) Nestin, marker of differentiated neural cells (F) β III tubulin, markers of glial astrocyte cells (G) vimentin, and (H) GFAP. The mean ± SD of three independent experiments are provided. *p < 0.05, **p < 0.01, ***p < 0.001.
Beside CD133 expression, we also investigated the MSC-CM impact on the expression of other markers associated with stemness of GSLCs, such as Sox-2, Nestin, and Notch-1. The qRT-PCR analyses revealed decreased expression of Sox-2 in NCH421k (1.5-fold, 1.8-fold), NCH644 (2.0-fold, 1.1-fold), NIB26 (6.3-fold, 4.3-fold), and NIB50 cells (5.1-fold, 5.2-fold) exposed to UC-CM and BM-CM, respectively (Fig. 5C). Likewise, a decreased expression of Notch-1 was noted in NCH644 (1.7-fold, 1.1-fold), NIB26 (3.9-fold, 2.9-fold), and NIB50 cells (2.3-fold, 1.7-fold) when exposed to UC-CM and BM-CM, respectively, yet with no change observed NCH421k cells exposed to either type of MSC-CM (Fig. 5D). Also Nestin expression did not change in NCH421k cells exposed to any MSC-CM, whereas it increased in NCH644 cells exposed to BM-CM (2.9-fold). Contrarily, Nestin decreased in NIB26 (3.0-fold and 3.3-fold) and NIB50 cells (7.2-fold and 7.6-fold) exposed to UC-CM and BM-CM, respectively (Fig. 5E).
The expression of the differentiation marker, such as a neuronal marker β III tubulin, decreased in NCH421k cells (1.6-fold and 2.5-fold) exposed to UC-CM and BM-CM, respectively, and in NIB26 cells (2.2-fold) exposed to UC-CM only, whereas it increased in NCH644 cells exposed to UC-CM (2.6-fold) and BM-CM (5.2-fold) (Fig. 5F). In contrast, the expression of glial cell markers vimentin and GFAP increased in CM-cultured GSLCs. Vimentin expression increased in NCH421k cells (1.4-fold, 1.6-fold), NCH644 cells (2.0-fold, 2.6-fold), and NIB50 cells (10.2-fold, 9.5-fold), when exposed to UC-CM and BM-CM, respectively (Fig. 5G). The GFAP expression increased in NCH421k cells (36-fold, 164-fold), NCH644 cells (2,265-fold, 6,909-fold), and NIB50 cells (14-fold, 416-fold), when exposed to UC-CM and BM-CM, respectively (Fig. 5H), whereas in NIB26 cells the GFAP gene expression was under detection level. This implies the shift in differentiation potential of GSLCs toward the astrocyte phenotype possibly induced by the MSC-CM.
MSC-CM Increased Sensitivity of NCH421k and NCH644 Cells to Temozolomide (TMZ) and 5-Fluoro-Uracil (5-FU)
Since the inhibitory effect of MSC-CM on the GSLC cell cycle and their tendency to differentiate was evidenced, we hypothesized on MSC-CM impairing the resistance of NCH421k and NCH644 cells to chemotherapeutic agents TMZ and 5-FU. Indeed, a viability of NCH421k cells simultaneously treated with TMZ and both types of MSC-CM for 48 h decreased compared to NCH421k cells exposed to TMZ only, with the decrease being significant at 2.5 μM TMZ for UC-MSC (20.3%) and 2.5 and 1 mM TMZ for BM-CM 10.6% and 10.4%, respectively (Fig. 6A). Similarly, the viability of NCH644 cells decreased after 48 h of simultaneous treatment with either 500 μM or 1 mM TMZ and UC-CM (10.7% and 18.6%, respectively). The viability of NCH644 cells also decreased 17.7% after simultaneous treatment with 1 mM TMZ and BM-MSC (Fig. 6A), when compared to TMZ-treated cells only. The simultaneous exposure of NCH421k to 1 mM TMZ and MSC-CM for 72 h resulted in their decreased viability in UC-CM and BM-CM presence (16.2% and 18.2%, respectively) compared to TMZ-treated cells only. Likewise, exposure of NCH644 to 1 mM TMZ and MSC-CM resulted in their decreased viability in UC-CM and BM-CM for 27.9% and 18.5%, respectively (Fig. 6B).
MSC-CM decreased resistance of NCH421k and NCH644 cells to TMZ and 5-FU. Viability of NCH421k and NCH644 cells incubated in TMZ and both types of MSC-CM for (A) 48 h and (B) 72 h was determined by MTS assay. Influence of 5-FU and both types of MSC-CM on viability of NCH421k and NCH644 cells was evaluated using MTS assay after incubation for (C) 48 h and (D) 72 h. The mean ± SD of three independent experiments are provided. *p < 0.05, **p < 0.01, ***p < 0.001.
A simultaneous treatment of GSLCs with 5-FU (1 μM, 50 μM, 150 μM) and BM-CM for 48 h resulted in a mean 10% decrease in NCH421k cell viability, compared to 5-FU-treated cells only (Fig. 6C). The viability of NCH644 cells after simultaneous treatment with 5-FU (1 μM, 50 μM, 150 μM) and UC-CM for 48 h showed a decrease in mean of 18.4%, when compared to 5-FU-treated cells only (Fig. 6C). Likewise, a simultaneous exposure to BM-CM with 1 μM and 150 μM 5-FU resulted in decreased NCH644 cell viability (9.5% and 15.4%, respectively), compared to 5-FU-treated cells only (Fig. 6C). Moreover, a simultaneous treatment of NCH421k cells with 1 μM 5-FU and BM-CM for 72 h decreased their viability to 13.1%, when compared to only 5-FU-exposed cells (Fig. 6D). The reduced NCH644 cell survival after 72 h was observed after simultaneous exposure of 150 μM 5-FU with UC-CM and BM-CM (27.9% and 14.5%, respectively) (Fig. 6D), when compared to 5-FU-treated cells only.
In summary, the above data imply the ability of the MSC paracrine signals to lower the resistance threshold of the GSLCs to chemotherapeutics, that is, to increase the cytotoxicity effect of two very different DNA-damaging agents: TMZ and 5-FU.
Discussion
In our previous results, we managed to demonstrate the capacity of BM-MSCs to impair the growth of established GBM cell lines U87, U373, and U251 through the release of certain cytokines, which we have identified (32). The established GBM cell lines contain a minute level of GSLCs; thus, the aim of this study was to reveal the effect of the MSC-secreted factors on the GSLCs expressing CD133, a known stem cell marker. In this study, we analyzed the effect of the medium conditioned by human MSCs (MSC-CM) derived from bone marrow (BM-MSC) and umbilical cord (UC-MSC), on the behavior of CD133+ GSLC lines NCH421k, NCH644, NIB50, and NIB26. We showed that the exposure of NCH421k cells to BM-CM and UC-CM caused the inhibition of the cell cycle in G0/G1 phase via decreased expression of cyclin D1 in the four tested GSLC lines, NCH421k, NCH644, NIB26, and NIB50. This is consistent with the study of Jaio et al. (19), where injected UCB-MSCs have been shown to decrease cyclin D1 expression in the C6 xenograft-bearing mice in vivo. The capacity of the UC-CM has been likewise proven to induce cell cycle arrest in the G0/G1 phase in vitro in the U251 cell line (50).
Apoptosis and senescence represent possible cellular responses to harmful environmental stimuli. CD133+ GSLCs are generally more resistant to apoptosis compared to differentiated cancer cells, as they express higher levels of antiapoptotic genes Bcl-2, Bcl-XL, Flip, and IAPs than differentiated CD133- GBM cells (11, 28). It is thus likely that, under our experimental conditions, neither the paracrine factors of BM-MSCs nor UC-MSCs could trigger the apoptosis or cause cell death in GSLCs as revealed by annexin V/PI staining, due to either insufficient levels or absence of the proapoptotic factors in MSC-CM or reportedly higher resistance of GSLCs to apoptosis. Contrary, UCB-MSCs were shown to induce apoptosis in glioma cells and xenografts upon indirect and direct cell-to-cell contact (13, 17, 20), where possibly the active cellular cross talk stimulated the secretion of additional factors. Our results indicate that soluble factors of MSC-CM alone cannot initiate apoptosis in GSLCs.
To our knowledge, the presented results are the first to show an induction of senescence as a preferred response of GSLCs to secreted factors from naive BM-MSCs and UC-MSCs, as demonstrated by changed morphology of the GSLCs, their SA-β-Gal staining, and gene expression analyses. The latter revealed upregulation of ATM gene (ataxia telangiectasia mutated), which most likely induced p53 signaling pathway (38). Indeed, by ontology enrichments analysis in KEGG, p53 signaling (5) was identified as the key pathway of senescence induction in NCH421k cells. Senescence PCR arrays revealed upregulation of CD44, collagen, type I, α 1 (COL1A1), MORC family CW-type zinc finger 3 (MORC3), NADPH oxidase 4 (NOX4), CDKN1A (p21), insulin-like growth factor-binding protein 5 (IGFBP5), and serpin peptidase inhibitor, clade E (SERPINE1) genes and downregulation of insulin-like growth factor binding protein 3 (IGFBP3), CDKN2A (p16), Cbp/p300-interacting transactivator, with Glu/Asp-rich carboxy-terminal domain, 2 (CITED2), fibronectin 1 (FN1), and protein kinase C, Δ (PRKCD) genes, in CM-exposed NCH421k cells. We hypothesized that MSC-CM induced senescence in these GSLCs through the activation of DNA damage response (DDR) program, since ATM and NOX4 genes were there found increased. ATM plays a key role in DNA damage recognition by activating the DDR, which via p53, in turn, upregulates p21 (CDKN1A) (38), which we also found upregulated in GSLCs exposed to MSC-CM (Fig. 3B). NOX4 is known to additionally stabilize senescence (49) by activating p21 via ROS generation (36) and may have caused the detected MMP loss in NCH421k cells. Similarly, IFN-β-treated fibroblasts through ROS generation activate their ATM/p53 senescence pathway (30) that is known to proceed by either p21 or p16 signaling (38). The MORC3 gene is involved in p53 induction of osteosarcoma cell's senescence (43) and SERPINE1, a downstream p53 target (26), is known to stabilize IGFBP5 in senescent cells (22, 34).
Taken together, we propose the senescence in GSLCs to occur via different pathways, since in our hands Cyclin D1 downregulation in all GSLCs associated with p21 upregulation and p16 downregulation in NCH421k cells, and contrary in the other three GSLC lines (NCH644, NIB26, and NIB50) with simultaneous p16 and p21 upregulation. As known, p21 regulates the initial phases of the cell cycle arrest, and its expression decreases after senescence is achieved (12). In contrast, p16 was shown to play an essential role in the later phases during the maintenance of the cell cycle arrest in already senescent cells, causing their permanent cell cycle arrest in G0/G1 phase (12). Thus, the mechanism of senescence induction in GSLCs could base on signals engaging intertwined p53-p21-pRB and p16-pRB tumor-suppressor signaling pathways, where cyclin-dependent kinase inhibitors (CDKIIs) p21 and p16 may play a prominent role (14, 38).
In summary, all the above implies the involvement of the two powerful tumor-suppressor pathways on activation and progression of senescence in GSLCs, caused by the MSC-CM paracrine signals. Namely, the p53 signaling pathway leads to the activation of pRB via activation of either p21 or the p16 pathways, which can both activate the pRB independently (12, 38). The phenomenon is clinically relevant, as antisense-mediated decrease of cyclin D1 was reported to inhibit the tumor growth and increase its chemosensitivity (41).
An alternative to GSLC elimination is to induce their differentiation toward chemo/radiotherapy less resistant, but still malignant GBM cells (10). GSLCs are characterized by a variable expression of CD133/Prominin-1 isoforms (46) and several other neural stem cell markers (Nestin, Musashi-1, Sox-1/2) (48), which correlate with their increased chemoresistance in vitro (9, 16, 28, 44) and unfavorable patient prognosis (1, 29, 52). Here we observed miscellaneous effects of both types of MSC-CM on CD133 expression in GSLCs. In the NCH644 cells exposed to MSC-CM for 72 h, we noticed decreased CD133 protein and mRNA expression, whereas in the NCH421k cells we failed to demonstrate a decrease in CD133 mRNA and protein expression to occur even after prolonged 7-day exposure to MSC-CM. This discrepancy can be explained by previous studies (9, 46) demonstrating that the CD133 epitopes 1 and 2 could be downregulated independently, contributing differentially to the total CD133 protein and mRNA levels. Furthermore, we demonstrated downregulation of Sox-2 and Notch-1 genes in GSLCs exposed to MSC-CM accompanied by the massive upregulation of the glial marker vimentin and GFAP, the astrocyte markers. Recent data suggest a major role of Sox-2 in the maintenance of the undifferentiated state of GLSCs (3), and the inhibition of Notch-1 was shown to induce neuronal and astrocytic differentiation in GSLCs (39). Taken together, these data suggest that MSC-secreted factors initiate a differentiation process in GSLCs toward the astrocytic cell lineage.
The standard therapy of GBM involves chemotherapy with alkylating agent TMZ. It is well established that TMZ improves the overall survival of GBM patients lacking MGMT expression (25). In GSLCs, the TMZ fails to display cytotoxic effect through DNA nicks (resulting from O6-methylguanin), due to their active O6-methylguanine-DNA-methyltransferase (MGMT) (28). This higher resistance of GSLCs toward TMZ compared to GBM cells has indeed been demonstrated (28, 44). Yet in our hands, the MSC-CM of both types were shown to decrease the resistance of NCH421k and NCH644 cells by a yet unknown mechanism, possibly affecting O6-methylguanine-DNA-methyltransferase levels. Based on our previous findings on cytokine secretion of BM-MSCs (32), we may speculate that factors like interleukin-24 (IL-24), which is commonly secreted from MSCs (40, 51), may be responsible for sensitizing the above GSLCs, as IL-24 was shown to act synergistically with TMZ to override the TMZ resistance of melanoma cells (53).
The chemotherapeutic 5-FU gets transformed in the cells into various metabolites, which inhibit the activity of thymidine synthase and thus prevent DNA synthesis. Here we noticed 5-FU decreased the survival of NCH644 cells more than the survival of NCH421k cells after its simultaneous treatment with MSC-CM. This difference, as in the case of TMZ, could be possibly explained by the previously reported heterogeneity of the GSLCs (37). The observed decreased chemoresistance of NCH421k and NCH644 cells points to a plausible linkage with the noticed shift of GSLCs to a more differentiated state in MSC-CM presence. Namely, the inhibition of Notch-1 was proven to sensitize GSLCs to radiation treatment (48). In addition, GSLCs from tumors were shown to overexpress multidrug resistance (MDR) proteins that protect them against cytotoxic drugs killing differentiated tumor cells. Sox-2 was proven to cause drug resistance by inducing ABC transporters (18). Thus, it is possible that the paracrine factors of the MSC-CM interfere with the activity of those proteins.
Taken together, we speculate that MSC-secreted cytokines are responsible for increased TMZ and 5-FU sensitivity of GSLCs, where the underlying mechanisms are yet to be investigated. These data may lead to the identification of novel key regulators secreted by MSCs, which may be used to enhance the effect of anti-GBM therapy, because of their ability to both increase GSLC differentiation and reduce the resistance to chemotherapeutics, in particular to the TMZ. Along with the effects on senescence, the MSC-CM factors either acting alone or in synergy are presenting a new insight important for the development of more efficient anti-GSLC targeting strategies.
Footnotes
Acknowledgments
This work was supported by the European Program of Cross-border Cooperation Slovenia-Italy 2007-2013 on Identification of New Glioma Biomarkers as Potential Diagnostic and Therapeutic Targets GLIOMA and by Slovenian Research Agency (Project No. J1-02474 to T.T.L.). The study was approved by the National Ethics Committee (Doc. No.96/02/12). We thank Dr. Nataša Tul Mandič, Dr. Miomir Knežević, and Klavdija Strmšek for providing us umbilical cords. The authors declare no conflict of interest.