Glioblastoma (isocitrate dehydrogenase (IDH) wild type, World Health Organization (WHO) grade 4) is the most aggressive form of malignant brain cancer that inevitably recurs despite the standard therapy, which consists of surgical resection, radio- and chemotherapy [1]. Only approximately 7% of patients survive more than five years after the initial diagnosis. The constant therapeutic failures imply an incomplete knowledge about the molecular biology of glioblastoma and amplify the necessity of finding new treatment strategies [2]. Glioblastoma was the first cancer to be included in a large-scale genetic analysis by The Cancer Genome Atlas Project (2008) in which common genetic alterations were identified, but now, in order to extend this knowledge, recognizing the proteomic changes, which are the functional consequences of genomic modifications is very important [3, 4, 5]. The utility of proteomics for understanding biology of cancer and the pathogenesis of the central nervous system (CNS) diseases is currently one of the most dynamic research areas in medicine. Although it is difficult to standardize glioblastoma specific markers due to their heterogeneity, identifying their biological signatures by determining their protein profile is a high priority because it enables them to be identified and results in the creation of libraries that show potential clinical biomarkers [6]. Moreover, proteomic analyses can provide unique information based on monitoring the longitudinal response to a treatment or by enabling the design of personalized therapy. All of this is necessary because novel treatment strategies and options, which once seemed to be promising such as the oncogenic signaling pathways including Receptor Tyrosine Kinase (RTK)/Ras/Phosphoinositide 3-Kinase (PI3K) p53 and retinoblastoma protein (pRB) signaling pathways, the vascular endothelial growth factor (VEGF)-targeting monoclonal antibody Bevacizumab, the DNA alkylating agents—lomustine and carmustine implants, and the checkpoint blockade inhibitors, have thus far been underwhelming in treating glioblastoma multiforme (GBM) [7, 8, 9].

Proteomic research on cancer cell lines, especially using the patient-derived lines, is important for identifying their biological markers. Although these preclinical models do not perfectly mimic the tumor microenvironment, patient-specific alterations at the molecular level are retained to a high degree. Today, proteogenomic and metabolomic studies are conducted on cell line models to: (1) develop and screen for novel anticancer drugs, (2) identify any novel mechanisms that are associated with variations in the drug response, (3) understand the mechanisms of action for the drugs that act directly on tumor cells, and (4) predict the clinical response to the drugs [10, 11]. Despite investigations on the mechanisms of GBM invasion in various human samples and preclinical studies, the strong inter/intratumor heterogeneity remains in high demand in the search for novel panels of biomarkers that will better characterize the different invasive glioblastoma phenotypes [12].

Because our previous study, which was performed on commercial T98G and primary (HROG02, HROG17) human glioblastoma cell lines showed differences in the sensitivity of cells to Temozolomide (TMZ) that were induced by changes in the glucose and oxygen concentrations in a culture medium [13], we decided to investigate whether there are differences in the proteome profile of the analyzed glioblastoma lines and, if so, could they be correlated with this effect.

The dot-blot method is a simple technic that is used to detect proteins and may can be used in the first stage of a screening test. It makes it easier to start begin high-throughput research. We compared the effects of TMZ on the protein expression in the commercial T98G cell line, lines that had been derived from a patient who had not been treated after resection (HROG02), and a patient with a recurrent disease (HROG17), all the while looking for common and individual markers. Moreover, we attempted to identify (if one existed) a correlation between the changes that are observed in the protein expression and the cellular migration as a quantifiable feature of cancer malignancy.

Finally, for the first time, we conducted microscopic observations of the prolonged effect of TMZ on: the dynamics of confluence by using the wound healing test, the quality and morphology of the viable glioblastoma cells in all of the lines, after the dead cells that were the result of being exposed to TMZ were removed. This model was designed to mimic the clinical situation when a resistant cellular subpopulation still exists after TMZ administration.

Therefore, the aim of the study was to analysis the influence of temozolomide on changes in glioma-related protein expression, cell morphology, and migration in three cell cultures.

In this study, the profile of 84 proteins in three glioma (grade 4) lines were analyzed: the commercial T98G, the primary HROG02 (derived from a treatment-naive patient) and HROG17 (derived from the relapse lesion) after exposure to TMZ, the first line alkylating drug that is used in patients with GBM. Despite the identical histopathological diagnosis—glioblastoma (but of different status relapse/treatment), our proteomics analysis revealed a unique protein signature for each line. However, in addition to this strong plasticity, ten proteins were found ubiquitously in all of the cases both with and without treatment: FGF, Survivin, Axl, SPARC, u-plasminogen, HO-1, Osteopontin, MMP-2, Cathepsin S, and Cathepsin B. The main functions of these proteins are related to proliferation, migration, pluripotency, immunosuppression, intercellular interactions, and extracellular matrix degradation [16, 17, 18, 19].

Altered metabolism, which promotes tumor survival and drives recurrence, is one of the prominent malignancy features of glioblastoma. In a previous study, we showed a strong functional link between the oxygen concentration/glucose availability, TMZ sensitivity and the intensification of the malignant features of glioblastoma cells, and the relationship between serum glucose and the glioblastoma cell Ki-67 expression in retrospectively analyzed clinical samples [20]. The data of the current proteomic analysis similarly identified the expression of the proteins that are involved in metabolism and their modulation by TMZ in the models from treatment-naive patients (commercial T98G and primary HROG02), but a quite different picture emerged in the glioblastoma relapse model, HRGO17.

Glioblastoma is also termed a “glioneuronal” tumor due to the expression of the neuronal antigens on the surface of cells that make up its mass [21]. Neuronal markers are aberrantly expressed in glioblastoma cells. One of them is Enolase—an essential enzyme that is involved in the glycolysis-metabolism pathway and that is elevated in many cancers including glioblastoma [22]. Physiologically, this enzyme is found in neurons and neuroendocrine cells. A role in the adaptation to cellular stress that is induced by treatment, hypoxia or a lack of nutrients has been suggested [22]. Although the enolase expression of glioblastoma is associated with a shorter survival time, the more general role of neuronal markers in glioblastoma is still open to discussion and their relevance must still be unraveled [23]. According to a single study, glioblastoma is able to create electrical synapses with the surrounding neurons that help drive glioblastoma invasion. Furthermore, glioblastoma cells can be reprogrammed into different subtypes of neurons (with a 20–40% neuronal efficacy) while losing tumorigenic capacity at the same time [24]. This kind of ability to transition microglia into neuron-like cells was reported earlier by our team as a mode of action of imipramine, an antidepressant drug. Conversion of glioma cells into neuron-like cells might, therefore, represent a promising novel therapeutic strategy and could open a new chapter in the treatment of this highly devastating cancer [25, 26].

Therefore enolase, which is a multifunctional protein, might be important in maintaining the glioblastoma malignant potential. Further support for this hypothesis comes from a study that has shown that enolase is a critical element in the adaptation to cellular stress and that it interacts with cathepsins. Moreover, when taking the important role of enolase in tumor immunology into account, there might also be a link between the glioblastoma metabolism and immunological pathways [27]. As a result, the immunometabolic profile of glioblastoma supports the development of an immunosuppressive tumor microenvironment, which ultimately results in an immunologically “cold tumor”. Not only a glioblastoma relapse but cerebral metastases also remodel the brain tissue and, in doing so, trigger inflammatory responses. These, in turn, stimulate growth, especially in the case of glioblastoma migration and further invasion [28].

Osteopontin, u-plasminogen, HO1-HMOX1, Cathepsins, and MMP proteins were detected as the common proteins in all three of the analyzed glioblastoma cell lines both in the group of “control conditions” and in the group after “TMZ exposure”. As a result, their constant expression (independent of an exposure to TMZ) can generate the potential for the degradation of the extracellular matrix (especially Cathepsins, MMP) and thus favor a recurrence. However, the “top” proteins, which were dominant in the relapsed HROG17 glioblastoma profile seem to dot the i’s and cross the t’s when determining the invasiveness strength (Kallikrein 5 and Thrombospondin). Osteopontin (OPN) also seems to be an attractive target for the immunomodulation of microglia. In our analysis, although this protein was detected at a moderate level in all of the studied glioblastoma lines, TMZ exposure did not influence it [29]. Based on the Human Protein Atlas, it is known that the highest expression in 17 of the analyzed cancers was OPN [30]. This glycophosphoprotein modulates multiple mechanisms of tumor-mediated immune suppression and has a pivotal role at the crossroads of inflammation and tumor progression. OPN also plays an important role in tumor repair, the processes of remodeling the extracellular matrix after injury, and is also responsible for the protumorigenic reprogramming of TAM [31]. In experimental study, an OPN deficiency was associated with a reduced immune-suppressive activity of M2 macrophages and a decrease in OPN/CD44 signaling, which can lead to the promotion of aggressive tumor growth as a consequence of the enhancement of the cancer stem cell activity in the glioma perivascular niche [31]. Moreover, elevated OPN levels in patients have been shown to correlate with a poor prognosis [32].

In experimental models, the serine protease Kallikrein-5 has been shown to accelerate migration and increase the invasion of cancer cells without affecting their proliferation capacity [33]. Moreover, it has the ability to sever extracellular matrix components (Collagen type I, II, III, and IV, Fibronectin, and Laminin) and the adhesion molecules (Fibrinogen and Vitronectin); both of which are critical qualities for modulating tumor invasion and metastasis [34]. The study of several cancer types including glioblastoma have emphasized the importance of Thrombospondin in modulating the immune responses and vascularization. Moreover, this multifunctional glycoprotein has a close association with the activation of transforming growth factor beta (TGF-$\beta$), thus orchestrating antigen-presenting cells and initiating microtube formation, thereby not only promoting the cytoplasmic extensions that are necessary for cell-to-cell communication but also promoting tumor invasion [35].

The up-regulation of the above-mentioned proteins is in agreement with our microscopic observations. In the analysis of the cells that had survived the exposure to TMZ, significant differences in the morphology and dynamic growth of the cells originating from the three studied GBM lines were observed. In the relapse glioblastoma cells, HROG17, a cellular phenotype with a weak proliferation, there was a strong plasticity of the cell morphology and a significant elongation of the soma cells, thus indicating the search for cell-to-cell contacts. However, the T98G and HROG02 cells behaved in quite the opposite manner and reached confluence twice as quickly and represented a growth-oriented phenotype.

However, in the wound healing test, it was observed that in an assay that mimicked the situation that occurs after tumor resection and around the resection margins, the HROG17 relapse glioblastoma cell line healed the wound twice as quickly as the primary cell models. Moreover, the HROG17 cells presented themselves with a long soma and had a higher morphological plasticity. Morphologically, the cells resembled the resting M2 anti-inflammatory form of microglia (ameboid). The T98G and HROG02 cultures were dominated by round microglia cells with an active M1 proinflammatory phenotype.

Because the response to central nervous system (CNS) damage or glioblastoma surgery is reactive gliosis, angiogenesis, and inflammation, the explanation of the observed differences between the analyzed three glioblastoma lines could also concern the microglia and its critical role in gliomagenesis and recurrence. Resident microglia and macrophages, which are known as tumor-associated macrophages, account for 30% to 50% of the tumor mass, which builds an immunosuppressive glioblastoma microenvironment and prevents a higher infiltration of functional antitumoral T and Natural Killer (NK) cells [36]. Tumor resection stimulates microglia activation, which is characterized by changes in the morphology, polarization, proliferation, and migration toward the inflicted injury [19]. Glioma cells that remain after the initial resection have been shown to transform incoming microglia cells into tumor-promoting cells, despite their normal role as executors of the innate immune response and sensing and eliminating abnormal cells [19].

TMZ, the monofunctional DNA alkylating agent and drug of first choice in glioblastoma therapy exerted different and mostly individual effects on each of the analyzed glioblastoma lines. Possibly, the most interesting effect was the appearance of TIE-2 and thrombospondin in the recurrent HROG17 line, but not in the primary glioblastoma cell lines. These proteins can shape and convert the glioblastoma profile to a more malignant form through their roles in the communication between glioma cells and the cells of the tumor microenvironment, thereby inducing chemoresistance by increasing the expression of ATP-binding cassette transporters, intensifying immunosuppression, and, similarly, triggering an escape from the immune surveillance [37, 38]. Among the “top” 10 proteins that were commonly expressed in all of the lines, we found that TMZ modulated 5 of them in different directions, namely: FGF, Cathepsins, Cathepsin b, Survivin, Axl and Osteopontin. Although these proteins have a wide range of functions, all of them are engaged in the immune response and create favorable conditions for the cellular processes such as the degradation of type IV collagen and the extracellular matrix, chromosomal instability, self-renewal, impaired cell-to-cell interactions, motility and proliferation, or the cell cycle. These proteins are also closely associated with the hyperactivation of the mitogen-activated protein kinase (MAPK) kinase pathway, which probably stimulates glioblastoma malignancy and has recently been suggested to be involved in glioblastoma relapse [39].

The complexity of glioblastoma disease and the results of our analysis of the cancer-related protein profiles after exposure of three glioblastoma cell lines to TMZ, suggest the need to look more deeply into several issues that can shape the malignant character of glioblastoma. Dynamic changes in the proteins expression were observed before and after the exposure to TMZ, thus demonstrating the impossibility of identifying a single protein that is responsible for glioblastoma development. Since proteins work as an interconnected net, we suggest that comprehensive protein profiles and protein panels have a greater credibility than even the most promising results of experimental studies that are dedicated to single proteins. Therefore identification of five proteins common to all of the lines: Cathepsin b, FGF, Survivin, AXL, Osteopontin that were modulated by TMZ suggests their potential as therapeutic targets or biomarkers to monitoring therapy effects.

Data are available upon reasonable request by contacting the corresponding author.

Study design: AMBW analyzed and collected data; GM performed statistical analysis; GM, KSG performed data interpretation; CL and ML made cell lines available and helped in preparing manuscript; EO analyzed data. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

Not applicable.

We would like to express our sincere thanks to Mrs. Edyta Golińska, Mrs. Anna Rynkowska, Mrs. Amy Hamilton, and Mr. Alexander Jurkowitsch for permitting us to use the LeviCell equipment and for their support in executing and analyzing the cell separation experiment.

This work was supported by a Grant from the School of Medicine, Medical University of Silesia, Katowice, Poland (KNW-1-121/N/9/0).

All authors have reported no conflicts relevant to the contents of this paper to disclose. Katarzyna Stec-Grosman is consultant of Laboratorium Medyczne Synevo, the judgments in data interpretation and writing were not influenced by this relationship.