The chaperone system in glioblastoma multiforme and derived cell lines: diagnostic and mechanistic implications

Background: Glioblastoma multiforme (GBM) is the most common and malignant primary brain tumor in adults. Novel treatments are needed to counteract the molecular mechanisms of GBM growth and drug resistance. The chaperone system (CS) members are typically cytoprotective but some, termedHsp, can become pathogenic and participate in carcinogenesis, alongwith the vascular endothelial growth factor (VEGF), and we investigated them in GBM biopsies and derived cell lines. The objectives were to identify diagnostic-prognostic biomarkers and gather information for developing chaperonotherapy. Methods: Cell lines from GBMs were established, characterized (morphology, growth characteristics, and specific markers), and stored. Chaperones and angiogenic factors [Hsp10, Hsp27, Hsp60, Hsp70, Hsp90, FLT-1 (VEGFR-1), FLK1 (KDR, VEGFR-2), and FLT-4 (VEGFR-3)] were observed in cells by immunofluorescence while the chaperones were measured in tumor tissue by immunohistochemistry. Results: Four cell lines were derived from four different GBMs; the cells were spindle shaped or polygonal and grew at high rates as adherent monolayers or clusters without evidence of contact inhibition. The astrocyte-specific glial fibrillary acidic protein (GFAP); and the neuronal NSE, malignancyVIM, and proliferation PCNA, markers were determined. The cells expressed GFAP but no NSE, indicating that they were primary glioblastoma cell lines, with high levels of Hsp10, Hsp27, Hsp60, Hsp90, and Flk1; and low levels of Hsp70, Flt1, and Flt4. Conclusions: Four cell lines were established derived from four out of ten GBM tumors studied. The cell lines showed intense positivity for chaperones studied and factors connected to malignancy and the tumors showed increased levels of chaperones, making them potential diagnostic-prognostic biomarkers and targets for anti-cancer compounds.


Introduction
Glioblastoma multiforme (GBM) develops from astrocytes and is the most aggressive primary cancer in human brain. GBM rarely metastasizes, it induces death through invasion of normal brain tissue and by resisting therapies. The 2016 World Health Organization (WHO) Classification of CNS Tumors combines histological and molecular parameters to define various types of gliomas [1]. Diffuse glioblastomas are separated into two groups, with and without mutation of the isocitrate dehydrogenase (IDH) gene, namely, WHO grade IV with and without IDH mutation [2]. These two groups of gliomas have different histopathological features, modes of progression, and prognoses [3]. IDH mutation frequently occurs in secondary glioblastomas, which grow slowly and have a favorable prognosis. The wild-type IDH occurs in older patients in the absence of any lower-grade precursor and have a worse prognosis [4]. Prognosis and clinical progression also differ significantly within the groups, indicating that there are other factors still to be identified that distinguish tumor sub-groups. In addition, the location and size of glial tumors determine the clinical presentation, requiring personalized monitoring and treatment strategies.
The distinctive feature of GMB is the recurrence and short survival time after aggressive therapies. This is due to the "angiogenic switch", consisting of the formation of a dense network of vessels tortuous and hyperpermeable, with production of pro-angiogenic factors, resulting in the uncontrolled proliferation, infiltration, and progression of the tumor. Angiogenesis is driven by mechanisms that lead to activation of the vascular endothelial growth factor (VEGF), one of the most important regulators of tumor growth [5]. The critical role of VEGF in angiogenesis makes it an appealing target for therapeutics. Preclinical data and early clinical trials with agents targeting VEGF prolonged patient survival; however, acquired anti-medication resistance and incomplete VEGF blockade made clear that other more efficacious therapies for GBM would have to be developed [6,7].
New avenues for research and for developing novel tools for prognostication and treatment of GBM have been opened by progression in the knowledge of the chaperone system (CS) and its functions as a physiological entity in health and disease, including tumors. The CS of an organism, e.g., a human, is composed of molecular chaperones, chaperone co-factors, co-chaperones, and chaperone receptors and interactors [8]. Molecular chaperones, the chief members of the CS, are proteins classified according to molecular weight into groups, encompassing the following ranges (in kDa): ≤34; 35-54; 55-64; 65-80; 81-99; 100-199; and ≥200; within these groups are families of phylogenetically related molecules termed heat shock protein (Hsp), such as the Small Hsp (those with the alpha-crystallin motif), Hsp40/DnaJ, Hsp70/DnaK, and Hsp90 families; and others, e.g., the CCT (Chaperoningcontaining TCP1) family [8]. The canonical role of the CS is maintenance of protein homeostasis, but it also has non-canonical functions pertaining to immunity, inflammation, and cancer [9][10][11][12][13][14]. In mitochondria, Hsp10 works with Hsp60 in forming a protein-folding machine, which is essential for cell life and survival. In some types of cancer, cells overexpress Hsp10, which accumulates in the cytoplasm, probably involved in neoplastic transformation [15,16]. Hsp27 is another chaperone that is increased in various types of cancer [17][18][19]. In glioma cell lines, Hsp27 is involved in a malignant phenotype, including proliferation, migration, and invasion [20]. Hsp60 is overexpressed in high-grade gliomas, and interacts with Hsp90 and cofactors, modulating tumor growth by preventing apoptosis in vivo [21]. Upregulation of members of the Hsp70 family has been observed in many tumors and a positive correlation with the WHO tumor grade has been reported for glioma [22]. Hsp70 is involved in the cancer-cell response to hypoxia, as its chaperoning activity is required for the folding, stabilization, and nuclear translocation of HIF (hypoxia-inducible factor) [23]. Cancer cells use the Hsp90 chaperone machinery to protect mutated and overexpressed oncoproteins, e.g., HIF and VEGF, from misfolding and degradation favoring cancer-cell survival [24]. GBM contains hypoxia niches that provide an environment favorable to the maintenance of stemness of the Glioblastoma Stemlike Cell (GSC) and, Hsp90 has been identified in the cells of this tumoral region co-localizing with stem-cell markers and HIF. Hsp90 is also considered a key protein for GBM cell migration and invasiveness, as well as modulation of survival and apoptosis [25].
The data briefly discussed above and many others in the literature support the notion that the CS system is involved in carcinogenesis, both against and for cancer development. Although the CS system components, particularly the molecular chaperones, are typically cytoprotective, if abnormal they can cause diseases, the chaperonopathies [26]. Therefore, it is pertinent to ask about the role of chaperones in GBM as possible biomarkers useful for differential diagnosis, prognostication, and patient follow up. Furthermore, it is highly likely that by revealing the quantitative and distribution patterns of chaperones in GBM in relation to clinical manifestations, response to treatment, and other pathological parameters, will provide clues for developing efficacious treatments targeting the chaperones, namely chaperonotherapy [27,28]. Consequently, we have been studying the CS in various cancers and, in the research reported here, we conducted an immunohistochemical evaluation in GBM tissue samples of the chaperones Hsp10, Hsp27, Hsp60, Hsp70, and Hsp90. In parallel, we also studied other molecules pertinent to GBM biology, such as Flt1 (VEGFR-1), Flk1 (KDR, VEGFR-2), and Flt4 (VEGFR-3) in primary glioblastoma cell lines derived from patient biopsies. The immediate objective of this project is to reveal features useful for tumor identification and differential diagnosis, while the long-term goal is to accumulate information that could serve as a platform for the development of therapeutic means targeting the pathogenic chaperones.

Patients and tissue samples
The patients enrolled in this study underwent surgery for supratentorial, hemispheric GBM at the Neurosurgery Department, Policlinico P. Giaccone, University of Palermo. In all cases, total tumor removal was donenamely, a resection of the entire contrast-enhancing lesion confirmed by MRI. All patients gave their written informed consent to study participation. This study was approved by the Ethics Committee of University Hospital AUOP Paolo Giaccone of Palermo (number 11/2018) in accordance with current legislation and the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. The samples were collected between 2019 and 2021; tumor diagnosis was performed according to the fourth edition of the WHO classification of tumors of the central nervous system. In our study, 10 patients with glioblastoma were enrolled at the Policlinico P. Giaccone and tissue biopsy samples were obtained from nine of them. Cell lines were established from 4 of the 9 samples. The mean age of the patients was 57.75 years (ranging from 53 to 64) and their clinical characteristics such as anatomical site, isocitrate dehydrogenase (NADP(+)) 1 IDH1 wild type (WT) or mutant, Ki67, and synaptophysin are summarized in Table 1. IDH mutations, Ki67, p53 and synaptophysin were determined in all glioblastoma cases from the Institute of Pathological Anatomy. In addition, 10 non-pathological samples of cortex from age-matched subjects, were used for comparative analysis, obtained from U.O.S. Forensic Medicine, Policlinico P. Giaccone.

Immunohistochemistry
Immunohistochemical staining was performed using formalin-fixed paraffin-embedded blocks. Briefly, 4-µmthick tissue sections, obtained by microtome cutting, were deparaffinized in xylene and hydrated by immersing in a series of graded ethanol. Antigen retrieval was performed placing the sections in epitope retrieval solution (0.01 M citrate buffer, pH 6.0) for 8 min and immersed for 8 min in acetone at -20 • C to prevent the detachment of the sections from the slide. After washing the sections with PBS (phosphate buffered saline, pH 7.4) for 5 min at 22 • C, the immunohistochemical reaction was performed by the streptavidin-biotin complex method using a Histostain®-Plus Third Gen IHC Detection Kit (Life Technologies, Frederik, MD, USA; Cat. No. 85-9073). The sections were treated for 5 min with Peroxidase Quenching Solution (reagent A of Histostain®-Plus 3rd Gen IHC Detection Kit, Invitrogen) to inhibit endogenous peroxidase activity. Endogenous peroxidase was inhibited by immersing the sections in 0.3% hydrogen peroxide for 10 min. Sections were then incubated with the primary antibody reported in Table 2. The immunohistochemistry procedure was performed as described previously [29]. Appropriate negative controls were run concurrently for each reaction. Nuclear counterstaining was performed using hematoxylin (Hematoxylin aqueous formula, nREF 05-06012/LN. Cat. No. S2020, Bio-Optica, Milano, Italy). Finally, the sections were observed with an optical microscope (Microscope Axioscope 5/7 KMAT, Carl Zeiss, Oberkochen, Germany) connected to a digital camera (Microscopy Camera Axiocam 208 color, Carl Zeiss, Oberkochen, Germany). All observations were performed by two independent observers (F.C. and F.R.). The percentage of positive cells was calculated in a high-power field (HPF) (magnification 400×) and repeated for 10 HPFs. The arithmetic mean ± standard deviation of counts was used for statistical analysis.

Establishment of cell lines
Tumor tissue was collected during resection and transferred cooled to the laboratory. Under sterile conditions, it was rinsed in PBS and, subsequently, obvious vessels, coagulated blood, and necrotic tissue were removed. Each tumor specimen was dissociated into small pieces, using scalpels, and dissociated further enzymatically by incubation with Collagenase Type II (250 U/mL, Sigma Aldrich, St. Louis, MO, USA), 0.25% of trypsin (Thermo Fisher, Waltham, MA, USA) in PBS at 37 • C for 18 h under shaking. At the end the incubation period, trypsin activation was stopped by adding medium containing 10% fetal bovine serum (FBS, Thermo Fisher Scientific). Centrifugation at 300× g for 5 min at 4 • C was performed to recover the cells from the enzyme solution and resuspended in Dulbecco's Modified Eagle's Medium (DMEM 1X, Thermo Fisher Scientific), supplemented with 10% FBS, and then the cells were filtered with a 70 µm cell strainer. The cells were washed by centrifugation and seeded in cell culture flasks with a medium composed of DMEM, 10% FBS, 100 U/mL penicillin (Thermo Fisher Scientific), and 0.1 mg/mL strep- tomycin (Thermo Fisher Scientific) at 5% CO 2 atmosphere at 37 • C. For suspension cultures and adherent cultures, T25 cell culture flasks were used. After a 1-day resting phase, the culture supernatant was removed, the cells were washed with PBS 1X (if there was cellular debris in the culture), and new culture medium was added. Change of culture medium and microscopic monitoring (phase contrast) were performed every 2 days. Aliquots of the cell lines have been stored frozen for future studies.

Hematoxylin-Eosin staining
Five thousand cells per well were plated in chamber slides and allowed to attach and proliferate for 24 or 48 h. After fixation (ice-cold methanol for 30 min, Carl Roth, Karlsruhe, Germany) the cells were stained with Hematoxylin-Eosin (H&E), using stain in hematoxylin solution (Papanicolaou Harris Hematoxylin, Bio-Optica: code: 05-12011) for 1 h at 22 • C; followed by counterstain with Eosin Y (Eosin Y aqueous solution 1%, Bio-Optica; code: 05-M10002) for 1.5 min. The slides were then mounted with histologic mounting medium and a cover glass, and finally observed with an optical microscope (Microscope Axioscope 5/7 KMAT, Carl Zeiss, Oberkochen, Germany) connected to a digital camera (Microscopy Camera Axiocam 208 color, Carl Zeiss).

Immunofluorescence and confocal microscopy
For immunofluorescence, the cells were placed in eight-well chamber slides, cultured for 24 h and fixed with ice cold methanol for 30 min. The fixed cells were washed with PBS pH 7.4 and then incubated with unmasking solution (trisodium citrate 10 mM, 0.05% Tween 20, pH 6) for 10 min at 22 • C. After rinsing twice with PBS, the cells were blocked with 3% (w/v) bovine serum albumin (BSA, Sigma Aldrich) in PBS for 30 min at 22 • C and incubated in a humidified chamber overnight at 4 • C with primary antibodies ( Table 2). The day after, the cells were washed twice in PBS and were incubated with a fluorescent secondary antibody (anti-rabbit IgG-FITC antibody produced in goat, F0382, Sigma-Aldrich, 1:100 dilution; anti-mouse IgG-TRITC antibody produced in goat, T5393, Sigma-Aldrich, 1:100 dilution; anti-rabbit IgG-Atto 647 antibody produced in goat, 40839, Sigma-Aldrich, 1:150 dilution; anti-mouse IgG-Atto 488 antibody produced in goat, 62197, Sigma-Aldrich, 1:150 dilution). The nuclei were counterstained with DAPI 33342 (1:1000, Sigma-Aldrich) for 15 min at 22 • C. Finally, the slides were covered with drops of PBS and mounted with coverslips. The images were captured using a Leica Confocal Microscope TCS SP8 (Leica Microsystems). Cell lines with cells immunopositive for GFAP and negative for NSE were defined as "primary glioblastoma cell lines".

Data analysis
Statistical analyses were carried out using the Graph-Pad Prism 4.0 package (GraphPad Inc., San Diego, CA, USA). Comparisons of immunohistochemical evaluations were made by using Student's t-test. All data are presented as the mean ± SD, and the level of statistical significance was set at p ≤ 0.05.

Culture efficiency
The tumor-tissue sample from one of the 10 patients did not pass the short-term primary stage of culture. Of the other nine glioblastoma-tissue samples, five did not attach after disaggregation and did not grow. The other four samples, however, did grow from the start and yielded selfreplicating passageable cell lines. This diversity of reaction to explantation probably reflects the area of the tumor from which the sample originates. If the sample is taken from a necrotic area, there are no cells, or extremely few that can grow in the cultures. If the tumor sample originates in a transitional area between necrosis and proliferation, cultivation is possible, but often large amounts of cell debris are present in the culture causing reducing cell viability still more. The presence of erythrocytes in the culture of samples with a high blood supply also leads to cell death. The four cell lines described here were designated GBM1, GBM2, GBM3, and GBM4. The time from explantation (p0) of the tumor samples to the first sub-cultivation step (p1) differed between the primary cell lines: one week for GBM2 and GBM3, and two and three weeks for GBM1 and GBM, respectively.

Morphological and growth characteristics of the four new cell lines
The new cell lines showed two types of growth: (a) monolayer with the cells growing side by side; and (b) fo-cal with the cells growing in clusters or islands connected by cell protrusions. The commercial cell line G166 grew as a monolayer. In slides stained with Hematoxylin-Eosin, the primary cell lines showed three different cellular shapes: polygonal (Fig. 3A,B), spindle or glia-like (Fig. 3C-F), and amorphous with cellular protrusions (Fig. 3G,H). Multinucleated giant cells were also present at low frequency. The G166 cell line showed a glia-like morphology (Fig. 3I,L). Aliquots of the cell lines have been stored frozen and they recovered and grew well after thawing.

Immunofluorescence: demonstration of GBM markers and mapping of chaperones in the new cell lines
To establish 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 resorted to immunofluorescence staining for markers used in glioma diagnosis with confocal microscopy, comparing our cell lines with the G166 line.
The primary cell lines were tested to determine the presence of GFAP, which is characteristic of astrocytes or glial cells [30]; Vimentin, a neuroepithelial precursor marker [31]; β-tubulin, a marker associated with tumor resistance to microtubule-targeting agents (MTAs) [32]; neuron-specific enolase (NSE), a neuronal marker [33]; and ALDH1, a stem-cell marker [34]. Positive immunoreactivities for GFAP, Vimentin, and β-tubulin, and negative immunoreactivity for NSE were detected in our four cell lines, as expected from GBM-derived cell lines (Fig. 4). The levels of GFAP varied being higher in GBM2, GBM3 and GBM4 compared to GBM1. The cell lines did not show staining for NSE except for GBM4, which showed a weak positive reaction. ALDH1 immunoreactivity was found in GBM1 (very weak), GBM2, and G166 cell lines, but the other cell lines were negative (Fig. 4). PCNA was present in all lines (Fig. 4). PCNA immunoreactivity was mostly confined to nuclei (diffuse or granular stained nuclei) in GBM1, GBM2 and GBM4 primary cell lines, whereas cytoplasmic staining was seen in GBM2, GBM3, GBM4, and G166 cell lines.
The immunopositivity for the molecular chaperone Hsp10 was observed in the cytoplasm of the primary cell lines (Fig. 5) and the cell line that showed the most intense reaction was GBM3, which derived from the youngest patient in the group studied. Hsp27 immunopositivity was intense in GBM2 (this line derived from the GBM patient with the shortest survival after surgery) and occurred also in GBM1, GBM3, GBM4 (derived from GBM patients with longer survival period after surgery) (Fig. 5). Hsp27 staining in the tumor tissue was not significantly different among the GBM patients. The Hsp60 reaction showed a granular and diffuse cytoplasmic positivity in our four GBM cell lines and in the G116 line (Fig. 5). The immunopositivity of Hsp70 was very low or absent in our four cell lines but it was present at very low levels in the G166 cell line (Fig. 5). These results obtained with immunofluorescence are in line with the immunohistochemical data (Figs. 1,2). Our four cell lines and G166 were positive for Hsp90 (Fig. 5). G166 cell line showed cytoplasmic positivity for all Hsps, including Hsp70 (Fig. 5).
Illustrative images of the immunofluorescence mapping of the VEGFR1 (Flt1), VEGR2 (Flk1), and VEGFR3 (Flt4) are shown in Fig. 6. The four GBM lines and the G166 cell line were positive for VEGFR-1 (Flt1) and for

Discussion
GBMs show a remarkable molecular heterogeneity that defeats the effectiveness of therapeutic approaches that are not patient specific. Understanding the cellular and molecular mechanisms of malignant growth in GBM will be instrumental for identifying new targets for treatment that could guide the development of personalized therapies. In this study, we focused on molecular chaperones. These are the chief members of the CS and are known to play key roles in carcinogenesis, both anti-and pro-tumoral roles [11,14,29,[35][36][37][38][39]. They are also promising candidates as biomarkers for differential diagnosis, prognostication, and patient follow up. We evaluated by immunohistochemistry and immunofluorescence the presence and subcellular localization of the chaperones Hsp10, Hsp27, Hsp60, Hsp70, and Hsp90 in biopsies of ten GBMs. Our experiments showed a high Hsp27 immunopositivity in the cytosol of all GBMs, by comparison with normal tissue. Hsp27 protein levels were higher in biopsies from GBM cases with short survival, a trend found also in other tumors, suggesting that elevated levels of Hsp27 may predict poor prognosis [40]. The immunopositivity for Hsp10 and Hsp60 was also elevated in the GBM tissue in contrast with the normal controls, with Hsp60 present in the cytoplasm, as shown for other types of solid tumors, in which the levels of this chaperonin correlated with tumor progression [15,41,42]. Hsp60 and Hsp10 are typically mitochondrial molecules, therefore, our finding them in the cytoplasm of GBM cells supports the notion that the chaperonins augment in the cy-tosol of tumor cells and are markers of malignancy. The chaperone Hsp70 have been found increased or decreased in human cancers by comparison with healthy control tissues [43][44][45]. An inverse correlation between Hsp70 levels and prognosis has been observed in some cancers [35,46]. In contrast to these findings, we did not detect a clear positivity reaction for Hsp70 in the GBM tissues we describe here, although we found low but still detectable levels of this chaperone in the commercially available stem line G166. This probably indicates that in this stem-cell line, Hsp70 is in the stem-cell nuclei, but it is absent in the other cell lines, in which stem cells have not been demonstrated. Hsp90 was detected many years ago in GBM [47,48]. In our study, we observed a higher Hsp90 immunopositivity in tumor tissue compared to normal brain tissue. All the observations on the chaperones in the tumor and derived cell lines by immunohistochemistry were confirmed by immunofluorescence.
The four cell lines reported here show some morphological heterogeneity, but most cells have a spindle-or glialike shape and there are also few multinucleated giant cells. The proportion of these cell types remained unchanged during cultivation and successive passages. The morphological heterogeneity of our cell lines reflects that of the original GBM tumors, which is characteristic of them. The cell lines displayed the markers of neuroglia, for instance GFAP and VIM, indicating that the cells derive from transformed glial cells. Available information suggests that the expression of GFAP and VIM is mostly confined to glialderived tumors, and that VIM can potentially be a useful marker for distinguishing undifferentiated GFAP-negative glial tumors. Other markers we measured in the cell lines are NSE, ALDH1, and PCNA. We observed no reaction for NSE, a low positivity for ALDH1, and positivity for PCNA. The absence of NSE, a neuronal marker, confirms the non-neuronal origin of the cell lines. ALDH1 has been suggested as a novel stem-cell marker in a small series of human glioblastomas [34], but the expression and functions of ALDH1 in the central nervous system (CNS) and its tumors are largely unknown. Our experiments showed that the primary cell lines do not have stem cell characteristics in contrast with G166, which showed immunopositivity for ALDHI. The immunofluorescence results with anti-PCNA antibodies indicate that the cell lines do show the parameters of malignancy that PCNA represent.
We observed positivity for Hsp27 in the cytosol of the cells and a diffuse positivity for Hsp60 and Hsp10 in the cytoplasm of all cell lines. The positive immunofluorescence reaction of Hsp10 and Hsp60 in the cytosol of the cells matched results obtained by immunohistochemistry in the tumor tissues. The fact that these chaperones were elevated from the first passages of cell growth, and remained elevated during cultivation, suggests that they are implicated in the onset and progression of carcinogenic mechanisms. It is possible that the increase in the amounts of Hsp27, Hsp10, and Hsp60 reflects the exaggerated requirements of tumor cells for new proteins correctly folded caused by their accelerated metabolism and growth and replication rates. The prognostic value of Hsp90 levels depends on the tumor type [48]. Hsp90 is often increased and associated with poor prognosis in various tumors, including GBM [49][50][51][52][53][54]. We found high levels of Hsp90 in the GBM2 and GMB4 cell lines with a cytoplasmic localization, in all passages. High levels of Hsp90 may indicate poor prognosis in some cases of GBM, considering that this chaperone can inhibit apoptosis [55].
GBMs are also characterized by neo-angiogenesis and invasion, which is a histopathological hallmark of these cancers associated with a worse prognosis [56,57]. We therefore examined the expression of the VEGFRs belonging to a superfamily of RTK (receptor tyrosine kinase) and considered key mediators of angiogenesis [58]. VEGFRs are augmented in grade IV glioma vasculature and grade IV glioma cells [59]. Previous studies found that VEGF and its corresponding receptors VEGFRs are significant prognostic factors for overall survival in patients with glioma [60]. Our results demonstrated positive VGFR-1 and VGFR-2 reactions with cytosol localization in the cell lines and a lower positive reaction for VEGFR-3 in the tumor biopsies. Previ- ous studies have shown that in normal brain a low or undetectable endothelial expression of VEGFR-2 can be found but in GBM it increases in parallel with tumor grade [61].
In summary, the levels of Hsp10, Hsp27, Hsp60, and Hsp90 were easily detectable in the GBM tumor tissue, even elevated some of them as compared with normal controls. In contrast, the levels of Hsp70 did not differ between the tumors and the normal tissue, being very low in all samples. These patterns of quantity and distribution of chaperones, if determined in many patients might, in the future, help in prognostication in GBM [60,62,63].
Our data provide a platform to elaborate a working hypothesis for future research, considering current knowledge about the CS and its roles in health and disease, including cancer. Hsp90 is a chaperone with a range of client proteins, including oncogene products (e.g., p53, AKT, RAF, EGFR, and MEK) involved in growth, survival, and cell-cycle regulatory pathways [64,65]. Hsp90 is upregulated within the hypoxic niche of human glioblastomas and a high level of Hsp90 is linked to the upregulation of HIF targets and CSC (cancer stem-like cells) marker genes [25] (Fig. 7).
Tumors with elevated Hsp90 had increased levels of the HIF target gene VEGF and high levels of the glioma stem-cells markers CD133 and nestin, supporting the notion that Hsp90 is activated by the CSC microenvironment [25]. Hsp90 and Hsp70 interact and regulate various tran-scription factors, signaling molecules, and kinases that are related to cancer, including NF-κB, p53, and others [66] (Fig. 7, Ref. [67]). Hsp70 is induced by HSF1, which is the main factor involved in the transcription of Hsp70 [68]. A pathway that is upregulated during tumorigenesis is the HIF/VEGF signaling axis: tumor hypoxia and other stimuli induce HIF expression and subsequent activity, leading to angiogenesis [69]. Mediators of this pathway include HIF and VEGFRs [70], which are regulated by Hsp90 that binds to the c-terminal tail of VEGFR-2 [71]. Hsp90 also modulates downstream effectors of VEGF-dependent signaling pathways [72]. Ligand-induced homodimerization of VEGFR-2 leads to autophosphorylation of VEGFR-2 on tyrosine residues, which drives activation of major signaling pathways that include ERK, mitogen-activated protein kinases (MAP kinases) [73], and stress-activated protein kinase (e.g., SAPK2/p38) [74]. Once activated, the stressactivated protein kinase triggers phosphorylation of Hsp27, which initiates actin remodeling and actin-based motility [75] (Fig. 7). Hsp60 could enhance tumor progression by maintaining mitochondrial homeostasis and activating the mTOR signaling pathway-overactive in GBM-which increases tumor resistance to treatment [76]. We therefore hypothesize that by interfering with the activity of chaperones (negative chaperonotherapy [27,77], for example Hsp60 or Hsp90 it may be possible to stop GBM growth. This strat- working hypothesis to be tested in search of specific targets, for instance chaperones, for anti-cancer compounds. Left panel; GBM cell. Hsp90 is at the center of pathways to maintain protein homeostasis, including interaction with protein degradations machineries that remove misfolded or otherwise defective proteins and generate building blocks, i.e., amino acids, for the new ones required in largerthan-normal amounts by cancer cells. Hsp90 also participates in tumorigenesis via the HIF/VEGF pathway. For example, hypoxia in the tumor mass induces HIF expression and activity, leading to VEGF production and angiogenesis. The chaperonins Hsp60 and Hsp10 are crucial for maintenance of the mitochondrial functional proteome and, thereby, ensure provision of energy to satisfy the needs of the tumor cell. In addition, Hsp70 and Hsp27 upregulate HIF activity increasing VEGF secretion. It is very likely that the CS, including the chaperones just mentioned, is key for chaperoning the protein molecules mediating the various pathways that are so important to satisfy the "greed" of the tumor cell for energy and proteins. The chaperones assist the tumor-cell proteins to mature and acquire a fully functional conformation and help them to translocate to the place in which they are to function. Therefore, components of the CS may be selected as targets for inhibition, namely negative chaperonotherapy, to hinder tumor survival and stop tumor growth and spread. Right panel; vascular endothelium cell. In vascular endothelial cells, Hsp90 interacts with members of the VEGF-regulated signaling pathway, including VEGFR2. A Hsp90/VEGFR2 complex exists in unstimulated endothelial cells [67]. However, binding of Hsp90 to VEGFR2 increases upon stimulation with VEGF, thus stimulating more angiogenesis. VEGF/VEGFR-2 promotes angiogenesis via a paracrine loop between cancer cells and vascular endothelial cells nearby. In addition, VEGFR-2 in vascular endothelial cells support tumor growth directly or through downstream activation of other signaling pathways, including the SAPK2/p38 pathway. The latter, once activated, triggers into action downstream molecules, such as Hsp27, which involves phosphorylation and stimulating actin remodeling and actin-based motility both necessary for angiogenesis. Hsp90 inhibition could alter the endothelial response and diminish or block angiogenesis by stimulating VEGFR2 proteolysis, providing another argument in favor of the use Hsp90 inhibitors as anti-cancer agents in GBM. egy could be applied in combination with the standard treatment options such as surgery, irradiation, and chemotherapy.

Conclusions
In conclusion, GBM is a tumor that is still little known in many ways, very aggressive and resistant to current therapies. It is necessary to find new therapeutic strategies that can improve treatment and follow-up tools. New avenues for research on GBM have been opened by progression in the knowledge of the chaperone system (CS) and its functions as a physiological entity in health and disease, including tumors. In this study we conducted an immunomorphological evaluation in GBM tissue samples and cell lines of some molecular chaperones and some angiogenic factors to investigate on their role as biomarkers useful for differential diagnosis, prognostication, and patient follow up. Therefore, on the one hand we wanted to provide useful details on the biology of GBM, on the other hand to add information that could be useful for the development of new therapeutic strategies. Finally, considering the current knowledge on CS and its roles in health and neoplastic diseases, we have developed a working hypothesis to be tested in search of specific targets, for instance chaperones, for anti-cancer drug compounds.

Author contributions
GA, CCB, CC and FR designed the research study. GA, LP, RP and AP performed the research. RB and FR analyzed the data. FG, AMF, and AA provided the samples. GA, RB, FR, wrote the original draft of the manuscript. AJLM, ECdeM, FB and FC wrote, revised, and edited the manuscript. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript.

Ethics approval and consent to participate
This study was ethically approved by the Ethics Committee of University Hospital AUOP Paolo Giaccone of Palermo (number 11/2018) in accordance with current legislation and the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments.