Characterization of adherent primary cell lines from fresh human glioblastoma tissue, defining glial fibrillary acidic protein as a reliable marker in establishment of glioblastoma cell culture

Abstract Background Primary adherent glioblastoma cell lines are an important tool in investigating cellular and molecular tumor biology, as well as treatment options for patients. Aim The phenotypical and immunocytochemical characterization of primary cell lines from glioblastoma specimens during establishment is of great importance, in order to reliably identify these cell lines as primary glioblastoma cell lines. Methods and Results Sixteen primary adherent cell lines out of 34 glioblastoma samples (47%) were established and further characterized. For phenotypical characterization, morphology and growth characteristics of the cells were classified. The cell lines had a high growth rate with a doubling time of 2 to 14 days. Morphologically, the cells displayed spindle‐form or polygonal to amorphous shapes and grow as monolayer or in foci without evidence of contact inhibition. The cells were able to migrate and to form colonies. For further characterization, the protein expression of the astrocyte‐specific protein glial fibrillary acidic protein (GFAP), the glial marker S100B, the neuronal marker TUBB3, and malignancy marker VIM, as well as the progenitor markers NES and SOX2, the proliferation marker MKI67, and the fibroblast marker TE7 were determined. Based on the immunocytochemical validation criterion of a coexpression of GFAP and S100B, 15 out of these 16 cell lines (94%) were defined as primary glioblastoma cell lines (pGCL). All 15 pGCL expressed TUBB3 and VIM. NES and SOX2 were stained positively in 13/15 and 6/15 pGCL. MKI67 was expressed in 11/15 and TE7 in 2/15 pGCL. Conclusion These results point out that in self‐established primary adherent glioblastoma cell lines, the expression of the specific astrocytic and glial markers GFAP and S100B and of the malignancy and progenitor markers VIM, NES, and SOX2 has to be validated. These data show that primary cell lines of glioblastoma origin with high malignant potential are reliably to establish using standardized validation criteria.


| INTRODUCTION
Malignant gliomas are the most common and most aggressive brain tumors in adults with currently no cure, 1 since they are characterized by pronounced invasiveness, as well as extensive intra-and intertumor heterogeneity 2-5 and a distinctive chemo-and radio-resistance. The standard treatment regime covers surgery to remove as much of the tumor as possible, followed by combined radio-chemotherapy. 6 New data show an additional benefit of an additive effect of alternating electric fields. 7 Given the overall poor prognosis, there is still a need to develop improved therapeutic options.
Reliably identified and characterized glioblastoma cell lines as a sustainable source of vital and proliferating malignant cells are invaluable for investigating tumor biology or for functional analyses such as response prediction. 8,9 Cancer cell lines, as the primary in vitro model system, are the standard both for exploring the basic cellular and molecular tumor biology and as preclinical models for testing new treatment modalities. For this reason, there is still a high demand for rapidly available cell models of primary brain tumors that are very close to the original tumor. Cell lines, which have been used widely in neuro-oncological research, are adherent glioblastoma cell lines established around 40 years ago. [10][11][12][13][14] It has become increasingly clear, however, that phenotypic characteristics and the multitude of genetic aberrations found within repeatedly in vitro passaged cancer cell lines often bear little resemblance to those found within the corresponding primary human tumor. 15 Primary cell lines directly derived from operations and then cultivated over as few passages as possible, more closely mirror the primary tumor, than commercially available cell lines do. Adherent primary glioblastoma cell lines provide a versatile and renewable resource to analyze the biology of tumor cells. Tumor cell proliferation, cell death and migration represent potential therapeutic opportunities that are accessible in adherent primary glioblastoma cell lines, and one can screen for agents that selectively and directly target them. This led us to cultivate biologically relevant adherent primary glioblastoma cell lines to define standardized characterization criteria for the validation of these cell lines. We are routinely able to generate characterized adherent primary glioblastoma cell lines. Continuous adaptation of our current technique has improved the processes for cultivation of cell lines directly from fresh glioblastoma tumor samples.
There is some literature available on the establishment of primary glioblastoma cell lines. Unfortunately, phenotypic as well as immunocytochemical characterization using a panel of marker proteins to confirm the identity of established tumor cells is often missing. Often only single primary glioblastoma cell lines are generated for specific analyses but not described in detail. Four work groups 14,[16][17][18] described the establishment and analysis of a number of primary glioblastoma cell lines. However, they did not prove the expression of specific glial markers for these cell lines. Only Mullins et al, 18 who established 17 primary glioblastoma cell lines, analyzed the marker expression of glial fibrillary acidic protein (GFAP), NES, and VIM in five cell lines using flow cytometry as an example. Among these cell lines, 75% to 95% GFAP-expressing cells could be detected.
This lack of reliable characterization of self-established primary glioblastoma cell lines demonstrates the urgent need for an accurate immunocytochemical analysis during cell line establishment. There are several well-established markers used for diagnosis of glioblastoma, but none of them is solely tumor specific. This limiting factor in the investigation of glioblastoma cells in vitro led us to investigate a panel of different markers for immunocytochemical analysis. This marker panel includes the astrocytic marker GFAP, the glial marker S100B (S100 calcium binding protein B), the neuronal marker TUBB3 (tubulin beta 3 class III), the tumor marker VIM (vimentin) and the progenitor markers NES (nestin) and SOX2 (SRY-Box 2), as well as the proliferation marker MKI67 (Ki-67) and the fibroblast marker TE7. The immunoprofile of the glial marker GFAP in glioblastomas is similar to astrocytomas. 19,20 Other, but less specific astrocytic, markers are S100B and MAP2a (microtubule-associated protein 2a). 21 Neuronal markers as beta III tubulin (TUBB3, Tuj-1), neurofilament protein (NFP), and neuron-specific enolase (NSE) are aberrantly expressed both in glioblastoma cell cultures and in patient biopsies. 22 The progenitor markers NES, a cytoskeletal protein expressed during the development of the central nervous system, and SOX2, a transcription factor responsible for maintaining stem cell features of embryonic stem cells and pluripotent stem cells, 16,23 are expressed under serumfree culture conditions but also in the first passages of primary cell lines. 20 Both have been found to be up-regulated in cancer, including high-grade gliomas. 23,24 They promote stem cell features, tumor cell proliferation, migration, and invasion. 20 As validation criterion, we have defined that all cell lines with an immunopositivity for both markers GFAP and S100B are primary glioblastoma cell lines (pGCL).   Doubling times were then calculated using an online calculator. 31

| Morphology assessment
Five thousand cells per well were plated in triplicate in 24-well cell culture plates and allowed to attach and proliferate for 48 hours. After fixation (ice-cold 70% ethanol for 15 minutes, Carl Roth, Karlsruhe, Germany) and staining with methylene blue solution (Carl Roth, Karlsruhe, Germany), growth type of the cell culture (monolayer, focal growth) and the cellular phenotype (dendritic-like, spindle-form, polygonal, amorphous, with or without cellular protrusions) was assessed to divide the cell cultures in subgroups according to their morphology.

| Senescence assay
Five thousand cells per well were plated in triplicate in 24-well cell culture plates and allowed to attach for 24 hours. After fixation, cells were stained overnight for senescence-associated ß-galactosidase (SA-ß-Gal) as described by Dimri et al. 32 Senescent cells were quantified by light microscopy. Only strong SA-ß-Gal positive cells were counted; it reflects the replicative age of cultivated cells. A proportion of up to 5% is usual for a proliferating cell culture.

| Migration analysis
The cells were plated in 24-well cell culture plates in triplicate and cultivated for 36 hours to achieve confluence. The cells were then synchronized for 12 hours in serum-free 1x DMEM. In the confluent cell layer, a straight line was scratched using a p10 pipet tip. The cells were incubated for further 24 hours, and images were acquired after 0, 6, 12, and 24 hours from the same field of view. The scratch distance was calculated using ImageJ 1.47v (NIH, Bethesda, Maryland 33 ) as mean of 10 horizontal lines drawn over the scratch. Furthermore, the cell cultures were divided into subgroups according to their migratory behavior: no migration, single cell migration, and collective migration.  low, ≤5% of cells; medium, >5% to ≤50% of cells; high, >50% of cells.

| Immunocytochemical staining
For each cell line, one cover slip per antibody was stained, completely examined and subjectively evaluated. Examples were photographed using an AxioPlan microscope system (Zeiss, Oberkochen, Germany).

All cell lines with cells immunopositive for both GFAP and S100B
were defined as "primary glioblastoma cell lines (pGCL)."

| Statistical analysis
All statistical analyses were carried out with the IBM SSPS Statistics v23 software (IBM, Armonk, New York). Statistical significance was defined as P ≤ .05. For the analysis of protein expression levels (immunocytochemistry), Spearman's rank correlation was used. Scratch assay results were statistically evaluated with variance analysis (ANOVA). To further test intergroup comparisons for statistical relevance, the Mann-Whitney U test was used.

| Cultivation efficiency and doubling time
Out of 34 glioblastoma specimens, we successfully established 16 glioblastoma-derived primary cell lines (pCL, 47%, Figure 1A and Table 1) with continuous proliferation that could be passaged regularly. Twelve of 34 glioblastoma tissue samples (35%) resulted in short-term primary cell cultures (pCC), which did not grow further after the first or second passaging and therefore could not be analyzed or stored. The remaining 6/34 tumor samples (18%) did not attach and proliferate after disaggregation (noncultivable, nc). It is assumed that the area of tumor from which the sample originates can cause the differences in cultivability. If the sample was taken in a necrotic area, it is not possible to cultivate cells. If the tumor samples originate from the transitional area between necrosis and proliferating tumor, cultivation is possible, but often such large amounts of cellular debris are found in the culture that the cells die. The presence of erythrocytes during the cultivation of samples with a high blood supply also leads to cell death. Methods for lysis of erythrocytes were not tested in this study.
Time from explantation (p0) of the tumor samples to the first subcultivation step (p1) differed between the primary cell lines (22 ± 13 days) and the short-term primary cell cultures (32 ± 23 days).
This difference was not significant (U-test, P = .17). The primary cell lines expanded continuously with a doubling time of approximately 2 to 14 days ( Figure 1B and Table 2

| Morphological classification and growth characteristics
Morphological classification of the established glioblastoma-derived primary cell lines was performed to identify the cytomorphological diversity expected in the glioblastoma "multiforme" cell lines. For this purpose, morphological criteria of the cell layer and the cell shape were recorded. In addition, colony formation and migratory behavior of the primary cell lines were investigated (Table 2).

| Immunocytochemical characterization
To specify staining criteria for the definition of primary glioblastoma cell lines and to ascertain whether the glioblastoma-derived primary cell lines have similarities to glioblastoma tissue, we undertook an immunocytochemical staining for markers used in glioma diagnosis (Table 2 and Figures 3 and 4). The marker expression was compared to the commercially available glioblastoma cell lines (cGCL) A-172, LN-229, and U-251 MG. Only U-251 MG was immunopositive for GFAP and, in U-251 MG and LN-229, an expression of S100B was detectable. A172 showed neither GFAP nor S100B. Fifteen out of the 16 primary cell lines (94%) were characterized, retained the expression of GFAP and S100B as reliable markers of astrocytic or glial cells, and were defined as primary glioblastoma cell lines (pGCL). Only in one primary cell line, no GFAP and S100B expression was detectable.
It had to be defined as nonglioblastoma cell line based on the validation criteria. However, the proportion of GFAP-expressing cells varied markedly in the primary glioblastoma cell lines (Figures 3 and 4A): in 3/15 pGCL, it was high (>50%), in 8/15 pGCL, medium, and in 4/15 pGCL, low (≤10%). S100B expression was higher compared to GFAP.
In 12/15 pGCL, it was high (>50%), in 2/15 pGCL, medium, and in F I G U R E 1 Cultivability of the 34 processed glioblastoma samples. A, Sixteen glioblastoma-derived primary cell lines (pCL, black) and 12 short-term primary cell cultures (pCC, gray) were established. Six tumor samples were noncultivable (nc, light gray). B, Doubling time of the primary glioblastoma-derived cell cultures. Five pCC were fast growing (<3 days), seven medium (3-7 days), and four pCC slow growing (>7 days), compared to the three cell lines (cGCL, mean 1.7 days). C-F, Cytomorphological characterization of the 16 pCC. C, Layer morphology (FG, focal growth; ML, monolayer). D, Cellular shape. E, Migratory behavior. F, Colony formation only one pGCL, low. With increasing GFAP expression, the S100B expression decreased ( Figure 4B, For immunocytochemical characterization, antibodies against glial markers, neuronal, and neural precursor markers, as well as malignancy markers, were selected (GFAP, S100B, TUBB3, VIM, NES, and SOX2). To identify the established primary cell lines as glioblastoma cell lines, an immunopositivity for both glial markers GFAP and S100B was defined. GFAP is a protein involved in the structure and function of the cytoskeleton and is commonly used as an astrocytes marker: its expression is increased following brain damage or during degeneration of the central nervous system. 48 In glioblastomas, this antigen is strongly expressed in the cytoplasm of the tumor cells. 20 Although F I G U R E 3 Immunocytochemical staining of pGCL. Examples for specific staining of GFAP, S100B, NES, and TE7 (green, FITC) in each GFAP expression group are given. Cell nuclei were counterstained with H33258 (blue). Magnification ×200. A, High GFAP expression group, B, medium GFAP expression group, and C, low GFAP expression group predominant among the water-soluble brain proteins, S100B is also found in a variety of other tissues. It is expressed in schwannomas, ependymomas, gliomas, and almost all benign and malignant melanomas and their metastases. 49,50 Since healthy cells of the brain parenchyma cannot be cultured under these cultivation conditions, GFAP and S100B can be used as markers for tumor cells in primary glioblastoma cell lines. The ubiquitous expression of VIM as a malignancy marker in the investigated pGCL, which identifies the cultured reported. 54 It was identified as an independent prognostic factor for high-grade glioma patients. 55 TUBB3 as a microtubule protein mainly expressed in cells of neuronal origin has been revealed as overexpressed in many cancers including gliomas. 56 In gliomas, TUBB3 expression seems to correlate with an increased malignancy, high proliferative rates, and poor prognosis. 57,58 All pGCL established in this study displayed a strong immunopositivity for VIM and TUBB3. In addition to the astrocytic markers GFAP and S100B, these two markers can be used to confirm the malignancy of established cell lines. 16,59,60 The

ACKNOWLEDGMENT
Open access funding enabled and organized by Projekt DEAL.

CONFLICT OF INTEREST
The authors declare no conflicts of interest. visualization; writing-original draft; writing-review and editing.

DATA AVAILABILITY STATEMENT
All data analyzed during this study are included in this published article. The raw datasets used for analysis in the current study are available from the corresponding author on reasonable request.