Background
Meningiomas are leptomeningeal neoplasms thought to originate from arachnoid membranes that form the cranial and spinal meninges [
1]. They account for the most World Health Organization (WHO) classified Central Nervous System (CNS) tumors in the USA [
2], and are highly reported worldwide [
3]. WHO classifies meningiomas into fifteen variants with grade I–III. Although grade III variants are most at risk for recurrence, a portion of grade I tumors reoccur (up to 20%) and currently there are very few predictive makers for such group [
4‐
6]. One of the first chromosomal mutation identified for meningiomas was the deletion of the long arm of chromosome 22 [
7]. In particular, a role of the
type 2 neurofibromatosis (NF2)/Merlin gene located on 22q12 was implicated [
8]. Genomic analysis of meningiomas tissues identified deregulations in the oncogenic genes
Phosphoinositide 3-
kinase (PIK), RAC-
alpha serine/threonine-
protein kinase 1 (AKT1), the G protein-
coupled receptor smoothened (SMO), TNF receptor-
associated factor 7 (TRAF7), and Kruppel-
like factor (KLF4) [
9‐
11].
Deregulated molecular pathways identified based on bulk tumor analyses, have been utilized to predict prognosis and help determine appropriate targeted therapy [
12]. However, despite all recent development in targeted therapy, surgery and radiation therapy remain typically the main methods of treatment for meningioma, even though both pose post-treatment challenges depending upon tumor location [
13]. For several CNS tumors, clinical trials that test for combinations of standard chemotherapeutic agents are still under progress. Cisplatin and etoposide are standard chemotherapeutic agents for many tumors. Both drugs alone, in combination with each other, or in combination with other drugs were found to be effective in adult and pediatric patients with low and high grade gliomas [
14‐
17]. These drugs are not primarily cell cycle affected, thus they could potentially be useful for the slow growing meningioma cells [
18]. Preclinical determination of the degree of resistance of meningiomas to either drugs and the identification of associated markers may prove valuable in improving tumor grading and prognosis [
19].
A crucial component for the development of targeted therapy is the availability of live tumor models [
20]. Early passaged cell cultures represent the “Central dogma” of patients’ tumors more faithfully than long term commercial cultures, yet very few attempts have been made to use such cells for pharmacological testing in meningiomas [
21]. Growth associated receptors and cancer stem cells (CSCs), cancer cells that express stem cell markers and are highly tolerant to adverse growth conditions, have been identified in stable meningioma cell lines [
22‐
31], even in cultures that have been sustained in a wide range of media conditions [
27,
32,
33]. However, identified meningioma CSCs in primary cell lines are not commonly associated with drug resistance, possibly because such studies require abundant number of cells which is often restricted by the size of excised tissues [
34].
Previously we published the gene expression profiles for meningioma patients tissues collected for our cohort [
35]. For this work, we focused on examining the properties of the corresponding meningioma primary cell lines in relation to biology, CSCs and drug resistance.
Methods
Tumor tissue RNA were isolated and processed for microarray and gene expression analysis as previously described [
35], and data was deposited at the NCBI’s Gene Expression Omnibus under accession number GSE77259. For gene expression analysis, based on differentially expressed probe sets, an unsupervised clustering was performed for Jed36_MN, Jed49_MN Jed04_MN, Jed18_MN, Jed34_MN, and Jed40_MN and visualized in a hierarchical clustering graph containing a colored heat map and adjunct dendrograms for displaying expression values and distance metrics between objects (samples or probe sets), respectively. Biological significance of expression data was interpreted by employing the core functional analysis workflow of the Ingenuity Pathways Analysis Software (IPA) (Ingenuity Systems, Redwood City, CA, USA). The Ingenuity Knowledge Base served as reference data set. Significance of relationships between data set molecules and functional frameworks, e.g. canonical pathways, provided by IPA was indicated by Fisher’s exact test p values.
Meningioma cell cultures initiation
Meningioma specimens collected between February 2013 and December 2015 were obtained within 30 min of tumor removal and distributed into both RNAlater solution (Life Technologies) for RNA/DNA extraction and cell culture using mechanical and enzymatic dissociation methods [
36]. For cell culture initiation, surgical specimens were minced, dissociated with a scalpel in Hank’s Balanced Salt Solution (HBSS) and further incubated with 1× enzyme solution in HBSS [0.2 mg/mL DNase type 1 (2000 U/mg), Sigma; 0.4 mg/mL collagenase type 1a (125 U/mg), Gibco; HBSS 10 mL (calcium and magnesium free), Invitrogen]. Samples were rotated for 15 min and then were centrifuged at 1000 rpm (180×
g) for 3 min at room temperature, re-suspended in DMEM-F12 (Gibco), 10% FBS (HyClone), 100 U/mL penicillin, and 100 µg/mL streptomycin, and maintained in standard humidified incubators at 5% CO
2. After 24 h, non-adherent cells were removed by washing with Phosphate Buffered Saline (PBS; HyClone) and the adherent cells were cultured until they reached confluence, then split (1:2) every 12–30 days dependent on the cell line. For each of tumors Jed62_MN and Jed79_MN, tissues were divided into four representative portions and each was placed into a 25 cm
2 tissue culture flasks containing a type of media (DMEM-F12 and 10% FBS, or DMEM high glucose concentrations of 4500 mg/L (Gibco) with 10% FBS, or DMEM high glucose concentrations of 4500 mg/L (Gibco) with 5% FBS, or DMEM low glucose concentrations of 1000 mg/L (Gibco) with 10% FBS).
Determination of cell morphology and trends
Cell cultures were observed for a period of 4 weeks with an inverted microscope (Leica DMI6000) to determine their morphologies. Twenty images of cells were taken at 10× magnifications each week using a digital microscope camera (Leica DFC425). The average number of counted cells per cell line exceeded 500 per week. The weekly morphology percentages were calculated as the total number of cells displaying a particular morphology, divided by the total number of cells counted. Morphological profiles for cells were confirmed by immunofluorescence staining with mouse anti-Vimentin (1:100, ab8978, abcam). All cell lines and images were reviewed independently at least twice to confirm counts.
Immunofluorescence staining
For cell lines, cells were seeded on chamber slides and fixed with 4% Paraformaldehyde (PFA). For in situ staining, fresh frozen tissues were cut at 4 μM sections, then fixed with 4% Paraformaldehyde (PFA). The antibodies used for standard immunostaining were rabbit anti-Caspase-3 (1:100, ab4051, abcam), mouse anti-CD133 (1:100, W6B3C1, Miltenyi), rabbit anti-Sox2 (1:200, 09-0024, Miltenyi), mouse anti Sox2 (1:100, ab75485, abcam), mouse anti Nestin (1:50, ab6142, abcam), rabbit anti-Ki67 (1:200, ab16667, abcam), mouse anti-Vimentin (1:100, ab8978, abcam), rabbit anti- Frizzled 9 (1:100, ab150515, abcam), mouse anti-BMI1 proto-oncogene, polycomb ring finger (BMI1) antibody (1:100, ab14389, abcam), and rabbit Anti-Anterior Gradient 2 antibody (1:100, ab76473, abcam). For secondary goat antibodies, 488 anti-mouse (1:300, ab150105, abcam) and 555 anti-Rabbit (1:700, ab150074, abcam) were used. Pictures were taken at 20× magnifications using Leica DMI6000 microscope and Leica DFC425 camera. Photos were edited in Photoshop 7.0 and signal levels were compared to negative controls of secondary only. T test or Chiχ2 values were retrieved using Statistical Package for the Social Sciences (SPSS) Graduate Pack 21.0.
Growth inhibition assays
To determine IC
50 values for cisplatin and etoposide, preliminary growth inhibition experiments were completed for three cell lines that had fast growth capability (Jed38_MN, Jed45_MN, Jed49_MN), (Additional file
1: Figure S1). Cells were plated in 96-well plates at 5000 cells per well supplemented with DMEM-F12 and 10% FBS and left to adhere overnight at 37 °C in 5% CO
2. Following attachment, cells were treated for 2 h with cisplatin (in Saline) or etoposide (in DMSO) at increasing concentrations of drug. Twelve days following treatment, cells were fixed with 4% paraformaldehyde (PFA) and stained with crystal violet. 10% acetic acid was added to cells and absorbance was measured at 590 nm. The survival fractions (the absorbance values of the test well expressed as a percentage of the untreated control) were calculated. All assays were performed in triplicate.
Discussion
We examined multiple bio-parameters for meningiomas through analysis of both eight tissue samples and 15 primary cell lines. Whole transcriptome microarray analysis for six tumor samples identified two groups of meningiomas with differential stem cell related pathways and a number of novel stem cell related biomarkers, including AGR2. Biological characterization of primary meningioma cell lines also sub-grouped cell lines into two main types; G type with predominantly glial like cells that grow slowly and were less viable, and NG type that had a mix of neuronal like and glial like cells, grow faster and showed invasive properties. In addition, NG type cells had a significantly higher percentage of cancer stem cells that express CD133+ Sox2+ or AGR2+ BMI1+, and were more tolerant to treatment with the chemotherapeutic agents cisplatin and etoposide. Importantly, drug treatment of NG type cells resulted in the enrichment of CD133+ Sox2+ or AGR2+ BMI1+ cells.
Gene expression profiling has been employed to grade, identify and characterize various sub-types within tumors. In this study, unsupervised hierarchical clustering of tumor samples divided tumors into two groups that were primarily different in their stem cell pathways. In addition, three of the top differentially up-regulated genes Reelin, Calbindin 1 and Anterior Gradient 2 Homolog, and two of the most down differentially regulated genes Cytochrome P450 and F-Box Protein 32 are associated with stemness [
37‐
41]. Unfortunately, it was not possible to analyze RNA for all tissues that generated cell lines due to lower achieved quality and often less available fresh tissue material for those tissues that generated cell lines. However, we compared RNA expression levels of AGR2 published by an independent study [
42]. Their data from 68 patients support our findings as AGR2 expression reported in GEO showed a significantly higher average expression levels in grade (II+) tumors/or tumors that reoccurred compared to grade I tumours (grade I: 104.285, grade II+/reoccurred: 988.182, T test P 0.021). Importantly, tissue immunofluorescence used for four tumors, supported gene expression results for finding differences in stemness expression. In addition, CSCs rich areas appear to be also rich in AGR2 expression. AGR2 has been shown to be up regulated in breast, prostate and pancreatic cancers. Its function has been associated with the suppression of p53 phosphorylation and it was shown to interact with metastatic associated proteins [
43]. However, its association with high grade meningioma is novel.
Most importantly, tumors that formed group 2, with low stem cell gene expression, generated G type cell lines that also had a lower expression of stem cell related markers, while tumors that showed high stem cell gene expression (including Jed45_MN) generated NG type cell lines that showed similar characteristics. This indicated that the generated primary cell lines do represent the essence of their corresponding tumors’ tissues, as previously shown with glioblastoma multiform tumors [
44]. Thus functional characteristics observed in the retrieved cell lines are likely to benefit translational research for meningiomas. Unfortunately a few limitation of these primary cell lines remain, including full knowledge of the biochemical and genetic changes that they undergo from the point of initiation, difficulties encountered when attempting to harvest a large number of cells for end-point analysis while preserving enough cells for further culturing, as well as difficulties in establishing xenografts models, especially for the slow growing tumors [
45]. Thus, further work is required to improve knowledge related to these features. In addition, it would be beneficial to increase the number of cell lines analysed in the future in order to verify analysis on a larger scale.
Clear biological features, including morphology which is influenced by the repository of biochemical interactions flowing through the cell signaling system [
46], separated the derived meningioma cell lines into two groups. Pleomorphic features and multilayers were observed in previous studies and interpretations of these features led to a classification of slow or fast growing types [
26,
27], however no association to stemness or drug resistance was made. The pleomorphic feature of NG type cell lines may be able to influence growth as cellular diversity could result in the production of a variety of growth factors that may promote proliferation. It is also possible that NG type cell lines have a high potential for cell division through selection upon retrieval of actively proliferating cells. It is worth noting that all of G type cell lines were derived from grade I tumors while four out of five NG type cell lines had come from tumor tissues observed to have a relatively high mitotic index count. It is also possible that the smaller bipolar neuronal-like cells may be mechanically more capable of division [
47]. Notably, NG type untreated cell lines had significantly less nuclear Caspase-3, which may have enabled inhibition of apoptosis and promoted proliferation [
48].
Previous work has shown the presence of cancer stem cells in meningioma cell lines [
28‐
31], but did not associated novel markers with growth dynamic. In culture, both NG type and G type cell lines had cells co-positive for stem cell markers, however, NG cell lines showed increased number of CD133+ Sox2+ and AGR2+ BMI1+ co-positive cells. It is likely that the high frequencies of CSCs in NG cell lines are important for tumor dynamic growth and concurs with the notion that CSCs are pluripotent and generate pleomorphic cells which are likely to form pleomorphic tumors. However, it is important to consider that unlike stem cells, the precise cycling nature of cancer stem cells is still debatable and appears to be micro-environmentally influenced [
49]. In addition, although the number of CD133+ Sox2+ cells was higher in NG type cell lines compared to the G type, an average of 34.6% of all NG type cells were positive, thus representing only a fraction of the total cell number, and could not necessary be solely accountable for the growth dynamic of this type of cell lines. Importantly, Nestin expression was relatively high in both types. However, Nestin positive cells that were not proliferating, were more frequently detected in G-type cells. As well as being detected in stem cells, Nestin expression has been reported in several differentiated cells including astrocytes, and its acknowledgement as an exclusive marker for CSCs is still debatable [
50‐
52]. Perhaps in meningioma, Nestin overexpression occurs as an early event in the process of CSCs generation. This event could either be followed by further deregulations in stem cell-genes, such as Sox2, which result in the development of aggressive cancer stem cells, or if no further deregulations of stem cell-genes occur then a state of cell differentiation-like is encouraged instead. This is consistent with the hypothesis that CSCs develop through clonal evolution and become more complex as tumors progress [
53]. Importantly, our data shows a novel association of AGR2 in meningioma stem cells. It is of interest to note that while staining Jed40_MN tissue revealed very few cells stained with AGR2, the corresponding cell line showed an average of 20% of cells that expressed AGR2. This suggests that for cell lines there might be a minimal threshold for AGR2 expression, perhaps to support survival under artificial/stressful conditions. Together, these results support a practice of stem-molecular staining for functional classification of high grade tumors and sub-typing of meningiomas.
Importantly, NG type cell lines showed a significantly higher tolerance of cisplatin and etoposide treatment compared with G type cell lines. Although etoposide has been traditionally thought of as a cell-cycle specific drug, recent data suggests a strong binding of the drug to chromatin and implicates a binding affinity for histone proteins, in particular histone 1, thus contributing to DNA damage regardless of the cell cycle status [
54,
55]. Notably, although untreated NG type cell lines had less cells that express nuclear Caspase-3, the difference in expression levels between the G and NG type cell lines was maintained following treatment. This finding highlights a possible role of apoptosis avoidance and promotes the notion of nuclear staining of Caspase-3 as a useful biomarker of resistance or susceptibility to cisplatin and/or etoposide. However Caspase-3 requires further pre-clinical validation in appropriate animal models and proof-of-concept in human clinical trials for meningiomas. Importantly, CD133+ Sox2+, Nestin+ ki67+ and AGR2+ BMI1+ cells survived following cisplatin or etoposide treatment of NG type cells. The frequencies of surviving cells that express the three combinations of markers were different, suggesting that similar to glioblastoma, heterogeneous populations of CSCs may exist in meningioma [
56]. Future experiments are required to investigate this possible diversity and the molecular basis of drug resistance by CSCs. A number of therapeutic strategies against CSCs are underway [
57].
In conclusion, analysis of patterns for cellular architecture, drug tolerance and associated resistance makers for primary cell lines in combination with genomic analysis of corresponding tissues of meningiomas, led to the functional sub grouping of meningiomas and the identification of novel stem cells related markers that appear to be associated with drug resistance and are likely to play a role in tumor aggressiveness. Importantly, this work highlights the need to use stem cells molecular markers and early in vitro analysis to classify functional variants of meningioma and predicts tumor aggressiveness and drug resistance.
Authors’ contributions
Cell lines morphological analysis experiments were done by HQ, NM, IK and DH. Tumor and cell lines processing and retrievals were set up by SB, MB, IK, HQ, MS, GD, AA, AC and DH. Histopathological classifications of tumors were done by FA, AJ and JA. Immunofluorescence analysis and drug treatment was done by DH. Transcriptome profiling was done by HS and MA. Conception and design was set by ACA, FG, SB, MB, HS, AC, AA and DH. SA, FG, HS, KS and DH wrote the manuscript and all authors critically revised. All authors read and approved the final manuscript.