433 research article Localization patterns of cathepsins K and X and their predictive value in glioblastoma Barbara Breznik12, Clara Limback34, Andrej Porcnik5, Andrej Blejec6, Miha Koprivnikar Krajnc1, Roman Bosnjak5, Janko Kos78, Cornell's J.F. Van Noorden19, Tamara T. Lah12 1 Department of Genetic Toxicology and Cancer Biology, National Institute of Biology, Ljubljana, Slovenia 2 Internationaml Postgraduate School Jozef Stefan, Ljubljana, Slovenia 3 Department of Histopathology, Charing Cross Hospital, Fulham Palace Road, London, UK 4 Faculty of Medicine, Imperial College London, South Kensington Campus, London, UK 5 Department of Neurosurgery, University Clinical Centre Ljubljana, Ljubljana, Slovenia 6 Department of Organisms and Ecosystems Research, National Institute of Biology, Ljubljana, Slovenia 7 Department of Pharmaceutical Biology, Faculty of Pharmacy, University of Ljubljana, Ljubljana, Slovenia 8 Department of Biotechnology, Jozef Stefan Institute, Ljubljana, Slovenia 9 Amsterdam UMC, Cancer Center Amsterdam, Department of Medical Biology at the Academic Medical Center, Amsterdam, The Netherlands Radiol Oncol 2018; 52(4): 433-442. Received 23 May 2018 Accepted 11 September 2018 Correspondence to: Barbara Breznik, Ph.D., Department of Genetic Toxicology and Cancer Biology, National Institute of Biology, Vecna pot 111, SI-1000 Ljubljana, Slovenia. Phone: +386 59 232 870; Fax: +386 59 232 715; E-mail: barbara.breznik@nib.si Disclosure: No potential conflicts of interest were disclosed. Background. Glioblastoma is a highly aggressive central nervous system neoplasm characterized by extensive infiltration of malignant cells into brain parenchyma, thus preventing complete tumor eradication. Cysteine cathepsins B, S, L and K are involved in cancer progression and are overexpressed in glioblastoma. We report here for the first time that cathepsin X mRNA and protein are also abundantly present in malignant glioma. Materials and methods. Gene expression of cathepsins K and X was analyzed using publically-available tran-scriptomic datasets and correlated with glioma grade and glioblastoma subtype. Kaplan-Maier survival analysis was performed to evaluate the predictive value of cathepsin K and X mRNA expression. Cathepsin protein expression was localized and semi-quantified in tumor tissues by immunohistochemistry. Results. Highest gene expression of cathepsins K and X was found in glioblastoma, in particular in the mesenchymal subtype. Overall, high mRNA expression of cathepsin X, but not that of cathepsin K, correlated with poor patients' survival. Cathepsin K and X proteins were abundantly and heterogeneously expressed in glioblastoma tissue. Immuno-labeling of cathepsins K and X was observed in areas of CD133-positive glioblastoma stem cells, localized around arterioles in their niches that also expressed SDF-1a and CD68. mRNA levels of both cathepsins K and X correlated with mRNA levels of markers of glioblastoma stem cells and their niches. Conclusions. The presence of both cathepsins in glioblastoma stem cell niche regions indicates their possible role in regulation of glioblastoma stem cell homing in their niches. The clinical relevance of this data needs to be elaborated in further prospective studies. Key words: cathepsins; glioblastoma; immunohistochemistry; patient survival; cancer stem cell niches Introduction Glioblastoma (GBM, WHO grade IV) is the most aggressive and also most common primary brain tumor.1 Despite present treatment strategies, such as surgical removal, radiotherapy and chemother- apy, only 5% of GBM patients survive 5 years and mean patient survival after diagnosis is approximately 1.4 years.2 These poor survival rates are mainly due to infiltrating type of growth of GBM cells into surrounding brain parenchyma, and extensive tumor heterogeneity.34 Radiol Oncol 2018; 52(4): 365-369. doi: 10.2478/raon-2018-0037 434 Breznik B et al. / Cathepsin K and X expression in glioblastoma The invasive spread of GBM cells is tightly associated with production and secretion of proteolytic enzymes5, including lysosomal cysteine cathepsins, belonging to the C1A family of papa-in-like proteases.67 The cysteine cathepsin family comprises of 11 proteases, sharing the same pro-teolytic mechanism, based on similar structural elements. However, these proteases have distinct conformations and catalytic activity (Figure S1) and are different with respect to their tissue and cellular distribution patterns and their physiological roles. Cathepsins play distinct roles in cancer progression, including invasion, the development of therapeutic resistance8 and apoptosis.910 Besides the hydrolysis of selective proteins, cysteine cath-epsins participate in proteolytic cascades, where one protease activates one or several others in sequences that finally regulate hydrolysis of peptide and protein substrates, which is called protease signaling.10 For example, secreted cathepsins can be considered as initiators of extracellular matrix (ECM)-degrading cascades during cell invasion, by cleaving and activating serine proteases, and modifying the tumor microenvironment by cleaving ECM proteins, shedding cell-cell adhesion molecules and processing relevant cytokines and growth factors to enhance tumor progression.81112 On the other hand, cathepsins may also possess tumor-suppressive roles13, depending on the cellular context, which emphasizes the importance of in vivo analysis to understand functions of cathepsins in GBM pathobiology.14 We have extensively investigated expression of cathepsins B, L an S at the mRNA and protein levels15,16 and found considerable differences in the correlation between expression of cathepsins B, L or S in specific end-points of GBM progression. For example, cathepsin B is involved in GBM cell invasion7, whereas the nuclear fraction of cathepsin L plays a role in apoptotic threshold regulation in GBM cells.910 Cathepsin S also contributes to GBM progression in vitro17, although its inhibition did not impair GBM cell inva-sion.7 Very little is known about the expression and the role of two other cathepsins, i.e. cathepsins K and X, in GBM progression. Cathepsin K has recently been identified as one of the most differentially-expressed proteases in GBM tissue and cell lines as compared to normal counterparts.18 Cathepsin K belongs to the cathepsin L-like cluster of the C1A family. Highly-positive charged basic residues in its structure allow allosteric accommodation of negatively-charged resident glycosaminoglycans, enabling formation of complexes with unique collagenolytical activity, and unwinding of triple helical collagens19, thereby participating in ECM degradation, in particular in bone metastasis.20 Cathepsin K as a monomer also degrades growth factors and chemokines, such as stromal-derived factor 1a (SDF-1a)2122, thereby indirectly affecting signaling pathways and migration. The structure and activity of cathepsin X (also called cathepsin Z) show several unique features that distinguish it from other cysteine cathepsins.23 Cathepsin X exhibits solely carboxypeptidase activity and is activated by other lysosomal endo-peptidases. Cathepsin X expression seems to be restricted to cells of the immune system and it regulates their proliferation, maturation, migration, adhesion, phagocytosis and signal transduction.24 Various molecular targets of cathepsin X exopepti-dase activity have been identified, including the |3-chain of integrin receptors, y-enolase, profilin-1, chemokine SDF-1a and others.2324 Furthermore, cathepsin X has been detected in the brain where it is localized in neurons, glial cells and ependy-mal cells.2325 Increased cathepsin X expression has been associated with various types of cancer, such as lung, colorectal and gastric cancers.26-28 Cathepsins have been reported to be involved in migration and self-renewal of tumor-initiating and therapy-resistant GBM stem cells (GSCs).2930 GSC stemness and malignancy are maintained in specific microenvironments, so-called GSC niches, where these cells are protected from the immune system and therapy.43132 The final goal of GSC niche targeting as a new anti-GBM approach is to disintegrate GSC niches to increase GBM therapeutic sensitivity.22 Since cathepsins are potent modifiers of the tumor microenvironment, we speculate that they may exert specific functions in GSC niches by modifying ECM and processing cytokines and growth factors. The aim of this study was to explore expression patterns of cathepsins K and X in GBM tissue at the mRNA and protein level. At the transcriptome level, the data were obtained from publicly-available databases to determine if there is any association with survival of GBM patients. At the protein level, the immunohistochemistry was performed on serial sections of 21 human GBM samples, focusing on cathepsin localization in peri-arteriolar GSC niches. Moreover, we aimed to find correlations between gene expression of cathepsins K and X and GSC niche markers in GBM tissue samples. Radiol Oncol 2018; 52(4): 433-442. 435 Breznik B et al. / Cathepsin K and X expression in glioblastoma Materials and methods Expression of cathepsins K and X at the mRNA level and correlations with glioma stages and survival of glioma patients Analysis of expression of cathepsin K and X mRNA in glioma tissue and its association with patient survival was performed using the publically-available GlioVis data portal.36-37 Briefly, two datasets (RNA-seq) were used, TCGA_LGGGBM (RNA-seq platform) to compare expression of cathepsins in glioma grades II-IV and TCGA_GBM (RNA-seq platform) for all the other analyses. Two hundred twenty-six (226) patients with grade II, 244 patients with grade III, 156 patients with GBM (grade IV) and 4 non-tumor patients were included in the query. All statistical analyses were performed in GlioVis data portal using pairwise comparisons between group levels with corrections for multiple testing (p-values with Bonferroni correction), and the log-rank test for Kaplan-Meier survival curve analysis. Correlation analyses (Pearson correlation coefficient, r) were performed between mRNA expression of cathepsins and the mRNA expression of GSC niche markers. Patients and brain tumor samples Paraffin-embedded tissue sections were obtained from biopsies of GBM patients, who were operated at the Department of Neurosurgery, University Clinical Centre of Ljubljana, Slovenia in the period 2013-2016. The study was approved by the National Medical Ethics Committee of the Republic of Slovenia (Approval No. 92/06/13). All procedures followed the Helsinki Declaration. Altogether, 21 patients with newly diagnosed glioblastoma (WHO grade IV) before radio and/or chemotherapy were included in this study. All patients gave informed consent to be included in the study. The histological diagnosis was established by standard protocols at the Institute of Pathology, Medical Faculty, University of Ljubljana. Diagnoses and tumor sample data are shown in Table S1. Immunohistochemistry Immunohistochemistry was performed on serial GBM paraffin sections. Five |om-thick sections of GBM biopsy samples were prepared according to routine procedures of the Institute of Pathology, University of Ljubljana. Paraffin sections were de-waxed in 100% xylene (3 min) and then rehydrated in 100%, 96%, 50% and 0% ethanol (in each ethanol dilution for 3 min). Heat-mediated antigen retrieval was achieved with sodium citrate buffer (pH 6.0). Blockage of endogenous peroxidase activity in the tissue was performed by incubation with 3% H2O2 in 100% methanol for 30 min at room temperature. To reduce non-specific background staining, sections were incubated with 10% goat or rabbit normal serum (Sigma) in phosphate-buffered saline containing 0.1% bovine serum albumin. After tapping off the serum-containing buffer, sections were incubated overnight at 4°C with primary antibodies: rabbit anti-SMA (1:200; Abcam, ab5694), mouse anti-CD133 (1:10; Miltenyi Biotec, W6B3C1), rabbit anti-SDF-1a (1:200; Abcam, ab9797), mouse anti-CD68 (1:50; Dako, EBM 11), rabbit anti-cathepsin K (1:200; Abcam, ab19027) and goat anti-cathepsin X (1:200; R&D Systems, AF934). This step was followed by incubation with the anti-mouse, anti-rabbit or anti-goat secondary horseradish peroxidase-conjugated antibodies (1:200; Dako) for 1 h. Protein expression was detected using DAB (Abcam) or AEC (Vector Laboratories) as peroxidase substrate, and hematox-ylin was used for counterstaining. The negative-control staining was performed in the absence of primary antibodies (Figure S2). All sections were analyzed by light microscopy and images were taken using a Ti Eclipse inverted microscope (Nikon) and NIS elements software (Nikon). Expression of cathepsins is presented as percentage (%) of immunostained areas of tumor sections. Twenty visual fields per tumor section (20x magnification) in non-necrotic areas were quantified using ImageJ software (https:// imagej.nih.gov/ij/) and the Immunohistochemistry Image Analysis Toolbox.6 38 39 Statistical analyses All statistical analyses were performed using R and GraphPad Prims 7. Overall survival of GBM patients was calculated from the date of surgery to the date of death or last follow-up. Survival analyses were estimated by Kaplan-Meier survival curves and these curves were compared with logrank tests. A p value of < 0.05 was considered to indicate statistically significant differences. Results Expression of cathepsins K and X at the mRNA level is upregulated in glioblastoma as compared with low grade gliomas and normal brain tissue Expression of cathepsins K and X in GBM was compared with expression in lower grade gliomas Radiol Oncol 2018; 52(4): 433-442. 436 Breznik B et al. / Cathepsin K and X expression in glioblastoma u I or © • ïîîif Histology ¿¡if .= 4043 V) o ~QqQ Subtype_Verhaak_2010 ® Jy W \ » © fh •I Grade Subtype tferhaak 2010 Histology FIGURE 1. mRNA expression of cathepsins K and X in gliomas of different grades and GBM subtypes. Public transcriptomic datasets were used as described in Materials and methods. Higher cathepsin K and X mRNA expression was found in GBM (grade IV glioma) versus grade II and III glioma (A, D), in GBM versus normal brain (B, E), and in the mesenchymal subtype of GBM versus the classical, proneural and neural subtypes, as classified according to Verhaak et al.39 (C, F). Boxplots show the distribution of mRNA expression (log2) in glioma grade II and III and GBM. Data were retrieved from GlioVis portal.37 The significance was set at p < 0.01 (**), p < 0.001 (***). (WHO II and III), as well as with expression in normal brain tissue in publically-available transcriptomic datasets (The Cancer Genom Atlas -TCGA) using the GlioVis data portal.3637 We found higher cathepsin mRNA expression in GBM (WHO grade IV) than in lower-grade gliomas (WHO grade II and III) (Figure 1A,D) and higher cathepsin K and X mRNA expression in GBM as compared to normal brain (Figure 1B,E). Furthermore, we observed overexpression of both cathepsins in the mesenchymal subtype (Figure 1C,F), which has been reported in the literature as being therapy-resistant and a more aggressive subtype of GBM.40 Expression of cathepsin K and X mRNA was lowest in the classical and the proneural GBM subtype (Figure 1C,F). FIGURE 2. Immunohistochemical staining of cathepsins K and X in serial paraffin-embedded GBM sections. The expression of cathepsins K and X in 21 GBM samples, quantified as percentage (%) of immunostained areas and box plots show the distribution of cathepsin expression in GBMs (A). Heterogeneous imunohistochemical staining was found in different parts of GBM samples for cathepsin K (B, D) and for cathepsin X (C, E). Inserts present high magnification images of GBM cells containing cathepsin K and X protein. Cathepsin K (F) and cathepsin X (G) expression was present in specific cells in normal brain tissue of GBM patients. These cells were not further identified. Immunohistochemical labelling of cathepsins was performed with DAB as substrate (brown color). Cell nuclei were stained using hematoxylin (blue/purple). Scale bar = 50 pm. Inter- and intratumoral heterogeneity of cathepsin K and X protein expression in GBM tissues To determine protein expression of cathepsins K and X in GBM, we performed semi-quantitative im-munohistochemistry on serial sections of 21 paraffin-embedded GBM samples. We observed that the expression of cathepsin X in GBM tissue was higher (38.0% of immunostained areas), as compared to that of cathepsin K (11.6% of immunostained areas) (Figure 2A). Protein expression of both cathepsins was heterogeneous across tumors as well as within the same tumor (Figure 2B-G). Cathepsin K and X proteins were detected in cancer cells and stromal cells of the tumor microenvironment (Figure 2B-G). Radiol Oncol 2018; 52(4): 433-442. 437 Breznik B et al. / Cathepsin K and X expression in glioblastoma Predictive value of mRNA expression of cathepsins K and X is GBM-subtype dependent In order to evaluate the predictive value of expression of cathepsins K and X at the mRNA level in GBM, we performed Kaplan-Maier survival analysis, using public GBM cDNA microarray datasets. Cathepsin K mRNA expression did not correlate with survival of all GBM patients (Figure 3A). When the GBMs were stratified into different GBM subtypes according to Verhaak et at.41, again no correlation of cathepsin K mRNA expression and patient survival was found (data not shown). In contrast, significant differences in survival were found in GBMs with different cathepsin X mRNA levels. Kaplan-Meier estimates of median patient survival was 11.8 months for high cathepsin X tumors and 14.9 months for low cathepsin X tumors (log-rank p = 0.027) (Figure 3B). This correlation was dependent on the GBM subtype, as high mRNA levels of cathepsin X correlated only with the shorter survival of GBM patients of the classical subtype (median survival being 10.8 months for cathepsin X high-expressing tumors and 15.8 months for cathepsin X low-expressing tumors, log-rank p = 0.049). In mesenchymal and proneural GBM subtypes, the cathepsin X mRNA expression did not have predictive value for patient survival (Figure 3C). ® £ 50% - CTSK high (n-7a, medianil 2.5 months] - CTSK lew (n=77. median=l4.7 months) _ jg run--' p=0.4042 Wilcoxon p=0-3S3& £ 50% ' CTSX high [n-76 mcdian-11.6 months) CTSX low (n-77. median=14 9 months) "Log-rank p=0.0266 'Wilcoxon p=0 0215 Survival time (months) Classical GBM subtype 'Log-rank p-0-0494 > "WilcoHon p-Q OOS4 ¿ ¡o* Survival time (months) Mesenchymal GBM subtype Proneural GBM subtype Log rank p SÖ-iS WplMMOh p-0 534 Survival time (months} Survival time [months) Survival lime (months) FIGURE 3. Kaplan-Meier survival curves of overall survival in relation to cathepsin K and X mRNA expression. Cathepsin K mRNA expression did not correlate with survival of all GBM patients (A), whereas patients with tumors expressing high cathepsin X mRNA levels exhibited poorer survival than patients with low cathepsin X mRNA expression (B). When we stratified GBMs in the different subtypes, cathepsin X mRNA expression had predictive value in the classical GBM subtype only (p = 0.049 by log-rank and p = 0.0084 by Wilcoxons' test) (C). Data for the Kaplan-Maier survival curves were obtained from GlioVis data portal.3637 GBM tumors were stratified into two groups, tumors with high and with low cathepsin K or X mRNA expression, using median values as cutoff. Cathepsins K and X in peri-arteriolar GSC niches Cathepsins K and X were not only present in GBM cells (Figure 2), but also in stromal cells, such as endothelial cells of the tumor vasculature (Figure 4). Furthermore, the expression of both cathepsins was FIGURE 4. Serial GBM sections immunohistochemically stained for peri-arteriolar GSC niche markers and cathepsins K and X. SMA-positive smooth muscle cells were present in the tunica media of the arteriolar wall as indicated by black arrows (A). CD133- and SDF-1a-positive cells (B, C) were present in the cellular layers adjacent to the tunica adventitia of the arteriole. CD68-positive macrophages and microglia were found in peri-arteriolar regions as indicated by black arrows (D). Cathepsins K (E) and X (F) were expressed in a CD133-, SDF-1a- and CD68-positive areas around the arteriole and their expression overlapped that of CD133-positive cells as indicated by white arrows (B, E, F). Cathepsins and SDF-1a were present in the endothelial cells of arterioles as indicated by black arrows (C, E, F). Immunohistochemical labelling of SMA (A) was performed with AEC as substrate (red color) and of the other proteins (B-F) with DAB as substrate (brown color). Cell nuclei were stained using hematoxylin (blue/ purple). a, lumen of arteriole; ta, tunica adventitia; tm, tunica media. Scale bar = 100 pm. ® IÍÉÉMIÉ " W ' ... ... ■sto^st*^ »js Radiol Oncol 2018; 52(4): 433-442. 438 Breznik B et al. / Cathepsin K and X expression in glioblastoma . • » f CTSK FIGURE 5. High-magnification images of serial GBM sections labelled for CD133, SDF-1a and cathepsins K and X. The region around the arteriole as shown in upper left corner of the Figure was magnified. CD133- (A), SDF-1a- (B), cathepsin K- (C) and cathepsin X- (D) positive cells were adjacent to the tunica adventitia of the arteriole. Immunohistochemical labelling of proteins was performed with DAB as substrate (brown color). Cell nuclei were stained using hematoxylin (blue/purple). a, lumen of arteriole; ta, tunica adventitia; tm, tunica media. The interrupted black line indicates the outer border of the tunica adventitia. Scale bar = 20 pm. found in special regions of GBM tumors adjacent to the tunica adventitia of arterioles, in peri-arteriolar GSC niches, as identified by immunohistochemical localization of CD133/Prominin1 as the most-widely used GSC marker, a-smooth muscle actin (SMA) as smooth muscle cell marker, and SDF-1a/ CXCL12, which have been shown to be present in GSC niches (Figure 4).22'34'35 Briefly, CD133 is localized on GSC plasma membranes42, SMA is specifically localized in smooth muscle cells in the tunica media of arteriolar and venular walls34,35, whereas SDF-1a, a chemotactic cytokine, is produced by endothelial cells and is involved in angiogenesis43 and in the retention of GSCs within their niches adjacent to the tunica adventitia of arterioles.2244 We also show here that macrophages and/or microglia cells were present in the regions of GSC niches around arterioles using the macrophage and mi-croglia marker CD68 (Figure 4A-D). Cathepsins K and X were localized in CD133-, SDF-1a- an CD68-positive regions around arteri-oles, defined by the tunica media containing SMA-positive smooth muscle cells (Figure 4). Cathepsins were present in astrocyte-like cells and endothelial cells (Figure 4) as well as extracellularly (Figure 5) in regions of GSC niches adjacent to the tunica ad-ventitia of arterioles (Figure 5). Overlap of CD133, cathepsin K and cathepsin X immunostaining in GSC niches was observed (Figure 4B,E,F). Higher magnification images of CD133, SDF-1a, cathepsin K and X staining are shown in Figure 5. Taken together, GSC marker CD133, cytokine SDF-1a and macrophage marker CD68 are localized around ar-terioles where cathepsins K and X are also present. Correlation between gene expression of cathepsin K and X, selected GSC markers and peri-arteriolar GSC niche markers The statistical correlation between gene expression levels of cathepsins K and X and gene expression levels of GSC markers and peri-arteriolar GSC niche markers was determined using public microarray datasets as describe in Materials and methods.36 37 We analyzed gene expression of GSC markers CD133/PROM1 and SOX2, endothelial cell marker CD31/PECAM1, a-smooth muscle actin SMA/ACTA2, chemotactic cytokine SDF-1a/ CXCL12 and its receptor CXCR4 and macrophage/ microglia markers CD68 and Iba1/AIF1. Cathepsin K mRNA expression correlated negatively with GSC marker SOX2 (Pearson's r = -0.38, *** p < 0.001) and positively with several GSC niche markers: PECAM1 (Pearson's r = 0.40, *** p < 0.001), ACTA2 (Pearson's r = 0.23, *** p < 0.001), CXCL12 (Pearson's r = 0.35, ** p < 0.01), CXCR4 (Pearson's r = 0.23, *** p < 0.001), CD68 (Pearson's r = 0.22, *** p < 0.001) and AIF1 (Pearson's r = 0.29, *** p < 0.001) (Figure 6). On the other hand, cathepsin X negatively correlated with GSC markers PROM1 (Pearson's r = -0.18, * p < 0.05) and SOX2 (Pearson's r = -0.23, *** p < 0.001). A strong positive correlation was observed between cathepsin X gene expression and the endothelial cell markers PECAM1 (Pearson's r = 0.49, *** p < 0.001) and ACTA2 (Pearson's r = 0.45, *** p < 0.001), CXCL12 (Pearson's r = 0.45, *** p < 0.001) and CXCR4 (Pearson's r = 0.46, *** p < 0.001). The strongest positive correlation was found for cath-epsin X and the macrophage and microglia marker CD68 (Pearson's r = 0.80, *** p < 0.001) and AIF1 (Pearson's r = 0.65, *** p < 0.001) (Figure 7). Taken Radiol Oncol 2018; 52(4): 433-442. 439 Breznik B et al. / Cathepsin K and X expression in glioblastoma together, gene expression of cathepsins K and X negatively correlated with gene expression of GSC markers and positively correlated with several markers of GSC niches. The positive correlation between gene expression of cathepsin X and GSC niche markers was much stronger than that of cathepsin K and GSC niche markers. Discussion Despite intensive research and the introduction of multimodal therapy with surgery, irradiation and chemotherapy, GBM patients survival has not significantly increased.45 A better understanding of GBM pathobiology and the discovery of cancer bi-omarkers with predictive value are thus crucial for the improvement of GBM treatment. Such markers may become targets for personalized therapy. In the search for new biomarkers, "omics" analyses, followed by molecular validation is the usual approach. Immunohistochemistry is the most common type of analysis in oncology, due to its relative simplicity in the assessment of biomarkers for diagnosis, prognosis and prediction of responses to therapy. Therefore, we attempted to assess the predictive value of two potential protease bio-markers, cysteine cathepsins K and X. The selection of these cathepsins was based on the fact that a related protease cathepsin B has been consistently found to be predictive at the mRNA and protein level in various types of cancer, including gliomas and GBM.646 High cathepsin B expression in the endothelial cells correlated with low survival rate of glioma patients and enables the identificaton of patients at higher risk in order to follow these patients more carefully or treat them more aggressively. Although cathepsin B is involved in cancer cell and endothelial cell migration and invasion15, it is unlikely candidate as selective treatment target because it is widely distributed in normal cells. The high cathepsin K expression as was found in the present study confirmed our previous findings on cathepsin K mRNA upregulation in GBM cells and tissues as compared to their normal counter-parts.18 However, we did not find any association of cathepsin K mRNA expression with survival of GBM patients, as was shown for various other types of cancers, such as lung carcinoma and squamous cell carcinoma.19 Similar results have been reported for the cysteine cathepsin L by Strojnik and co-workers47, showing high cathepsin L expression in astrocytomas and GBM, but these high levels were not predictive. We have also shown that cath- FIGURE 6. Correlation of microarray-based gene expression levels of cathepsin K and GSC niche markers. Cathepsin K did not correlate with GSC marker CD133/ PROM1 (A), but negatively correlated with GSC marker SOX2 (B) and positively correlated with GSC niche markers CD31/PECAM1 (C), a-smooth muscle actin SMA/ ACTA2 (D), chemotactic cytokine SDF-1a/CXCL12 (E) and its receptor CXCR4 (F), as well as with macrophage/microglia markers CD68 (G) and Iba1/AIF1 (H). Trend lines indicate linear regression estimates. Log2-transformed mRNA expression data were obtained via GlioVis portal. epsin K is expressed in normal brain cells as has been reported for all brain regions of wild type mice48 where it has been associated with neurobe-havioral disorders such as schizophrenia.49 At the protein level, cathepsin K was detected in vesicles of neuronal and non-neuronal cells throughout the mouse brain and its deficiency was associated with a marked decrease in differentiated astrocytes, indicating a possible role of cathepsin K in stem cell differentiation. Cathepsin X has been linked to cancer progression as its upregulation has been found in several types of cancer.2328 Previously, high cathepsin X levels in serum and cancer tissues of patients with colorectal cancer and hepatocellular carcinoma, respectively, have been associated with shorter overall survival.2850 In present study, we confirmed the same trend for cathepsin X mRNA levels in 156 GBM patients. When GBM were stratified according to subtype41, high cathepsin X expression in Radiol Oncol 2018; 52(4): 433-442. 440 Breznik B et al. / Cathepsin K and X expression in glioblastoma 11 IB CTSX ® y i, ■ 11 13 CTSX •y - jr.ls • u_ io- Si ' * 8- • M >*,», „„ \ 9 11 13 ) CTSX 11 13 CTSX FIGURE 7. Correlation of microarray-based gene expression levels of cathepsin X and GSC niche markers. Cathepsin X negatively correlated with GSC markers CD133/ PROM1 (A) and SOX2 (B) and positively correlated with GSC niche markers CD31/ PECAM1 (C), a-smooth muscle actin SMA/ACTA2 (D), chemotactic cytokine SDF-1a/CXCL12 (E) and its receptor CXCR4 (F), as well as with macrophage/microglia markers CD68 (G) and Iba1/AIF1 (H). Trend lines indicate linear regression estimates. Log2-transformed mRNA expression data were obtained via GlioVis portal. classical GBM subtype showed predictive potential for poor survival. Cathepsin X promotes tumor processes by enhancing epithelial-to-mesenchymal transition (EMT) and by cleaving integrin receptors and profilin 1.235152 Moreover, invasion-promoting functions of cancer cell- and stromal cell-derived cathepsin X are mediated via RGD motifs in the protease prodomain, that binds to integrins and the ECM.53 This is the first observation that cathepsin X gene expression is significantly higher in GBM than in lower-grade gliomas and in normal brain, which is in line with studies in other types of cancer, such as hepatocellular carcinoma and melanoma, revealing the correlation between cathepsin X overexpression in cancer tissue and advanced tumor stages.50 In addition to the presence of cathepsin X in cancer cells in GBM tissues, its high levels were detected in endothelial cells of the tumor vasculature as well, which implies the involvement of cathepsin X in GBM angiogenesis, as has been found for cathepsin B.46 Interestingly, highest cathepsin X expression was observed in the mesenchy-mal subtype of GBM, which represents the most aggressive and therapy-resistant GBM subtype.4041 As already mentioned, GBM is heterogeneous at the molecular and cellular levels. Cathepsin K and X expression exhibited high inter- and intra-tumoral heterogeneity, which may be explained by differential infiltration rates of stromal cells, such as macrophages, endothelial cells, fibroblasts and lymphocytes that are also important sources of proteases.11 In addition, GBM subtypes may significantly differ in their microenvironment, for example, the mesenchymal GBM subtype exhibits a higher immune cell infiltration than all other subtypes.4154 At present, the only organized multi-cellular structures where cathepsins K and X have been found to be clustered, are the peri-arterioral regions that function as GSC niches.2233-35 In these niches, GSCs are surrounded predominantly by endothelial cells, pericytes, smooth muscle cells, fibroblasts and macrophages.455 In particular, cross-talk between endothelial cells and cancer cells is crucial for GSC propagation within the niches.5556 Furthermore, the chemotactic cytokine SDF-1a and its receptors CXCR4 and CXCR7 are important for the retention and maintenance of GSCs in their niches as well as for their radiotherapy resistance.4457 We localized cathepsin K and X in peri-arteriolar regions, positive for GSC marker CD133, smooth muscle cell marker SMA, SDF-1a and macrophage marker CD68. Thus, we confirmed our previous observations that cathepsin K is present in GSC niche regions and proposed that it is involved in GSC trafficking in/out of niches by proteolytic processing and inactivation of SDF-1a.22 Interestingly, we found an inverse correlation between cathepsin K and X mRNA expression and mRNA expression of GSC markers CD133 and SOX2 in GBM tissues. This is in line with our previous data58, showing an inverse correlation between CD133 mRNA levels and mRNA levels and activity of cathepsins B, L and S in isolated primary CD133-positive versus CD133-negative GBM cells. On the other hand, a positive correlation was found between the expression of cathepsins K and X and that of other GSC niche markers. Based on these results and the fact that cathepsin K cleavage of SDF-1a inhibits its chemotactic activity towards CXCR4-positive GSCs, we speculate that cath-epsins K and X may both enhance GSC migration out of their niches. This is a similar scenario that was reported by Kollet et al.59 for the mobilization Radiol Oncol 2018; 52(4): 433-442. 441 Breznik B et al. / Cathepsin K and X expression in glioblastoma of hematopoietic progenitor cells from bone marrow niches. In conclusion, our data support the concept that cathepsins K and X are associated with glioma progression, as both are progressively upregulat-ed from low grade to high grade glioma GBM, as we demonstrated by their significantly increased mRNA expression. High expression of cathepsin X on the mRNA level had predictive value for the survival of GBM patients. Furthermore, we have shown that cathepsin K and X proteins clustered in regions of peri-arteriolar GSC niches, and their mRNA expression levels correlated with expression of niche markers, implying a role of these two cathepsins in maintaining/trafficking of GSCs in or out of niches. The results of this study have to be confirmed in a further prospective study. Acknowledgements This work was supported by the Slovenian Research Agency (ARRS), Program P1-0245 (TTL) and ARRS young researcher grant (BB), and the European Program of Cross-Border Cooperation for Slovenia-Italy Interreg TRANS-GLIOMA. Authors' contributions All authors revised the manuscript and approved the final version of the manuscript. Specifically, BB designed the study, performed experiments, analyzed and interpreted the data and wrote the manuscript; CL provided GBM sections and optimized the immunohistochemical experiments; MKK performed experiments and analyzed the data; AB contributed to statistical analyses and interpretation of the data; AP, JK, RB provided GBM samples/patients data/sections/antibodies, analyzed and interpreted the data; TTL and CJFVN conceived and designed the study, analyzed and interpreted the data and wrote the final version of the manuscript. References 1. Louis DN, Perry A, Reifenberger G, von Deimling A, Figarella-Branger D, Cavenee WK, et al. The 2016 World Health Organization classification of tumors of the central nervous system: a summary. Acta Neuropathol 2016; 131: 803-20. doi: 10.1007/s00401-016-1545-1 2. Czarnek N, Clark K, Peters KB, Mazurowski MA. Algorithmic three-dimensional analysis of tumor shape in MRI improves prognosis of survival in glioblastoma: a multi-institutional study. J Neurooncol 2017; 132: 55-62. doi: 10.1007/s11060-016-2359-7 3. Cuddapah VA, Robel S, Watkins S, Sontheimer H. A neurocentric perspective on glioma invasion. Nat Rev Neurosci 2014; 15: 455-65. doi: 10.1038/ nrn3765 4. Roos A, Ding Z, Loftus JC, Tran NL. Molecular and microenvironmental determinants of glioma stem-like cell survival and invasion. Front Oncol 2017; 7: 120. doi: 10.3389/fonc.2017.00120 5. Mentlein R, Hattermann K, Held-Feindt J. Lost in disruption: role of proteases in glioma invasion and progression. Biochim Biophys Acta 2012; 1825: 178-85. doi: 10.1016/j.bbcan.2011.12.001 6. Colin, C, Voutsinos-Porche B, Nanni I, Fina F, Metellus P, Intagliata D, et al. High expression of cathepsin B and plasminogen activator inhibitor type-1 are strong predictors of survival in glioblastomas. Acta Neuropathol 2009; 118: 745-54. doi: 10.1007/s00401-009-0592-2 7. Gole B, Huszthy PC, Popovič M, Jeruc J, Ardebili YS, Bjerkvig R, et al. The regulation of cysteine cathepsins and cystatins in human gliomas. Int J Cancer 2012; 131: 1779-89. doi: 10.1002/ijc.27453 8. Olson OC, Joyce JA. Cysteine cathepsin proteases: regulators of cancer progression and therapeutic response. Nat Rev Cancer 2015; 15: 712-29. doi: 10.1038/nrc4027 9. Kenig S, Frangež R, Pucer A, Lah T. Inhibition of cathepsin L lowers the apoptotic threshold of glioblastoma cells by up-regulating p53 and transcription of caspases 3 and 7. Apoptosis 2011; 16: 671-82. doi: 10.1007/ s10495-011-0600-6 10. Lankelma JM, Voorend DM, Barwari T, Koetsveld J, Van der Spek AH, De Porto AP, et al. Cathepsin L, target in cancer treatment? Life Sci 2010; 86: 225-33. doi: 10.1016/j.lfs.2009.11.016 11. Breznik B, Motaln H, Turnšek TL. Proteases and cytokines as mediators of interactions between cancer and stromal cells in tumours. Biol Chem 2017; 398: 709-19. doi: 10.1515/hsz-2016-0283 12. Kramer L, Turk D, Turk B. The future of cysteine cathepsins in disease management. Trends Pharmacol Sci 2017; 38: 873-98. doi: 10.1016/j. tips.2017.06.003 13. Lopez-Otin C, Matrisian LM. Emerging roles of proteases in tumour suppression. Nat Rev Cancer 2007; 7: 800-8. doi: 10.1038/nrc2228 14. Quail DF, Joyce JA. The microenvironmental landscape of brain tumors. Cancer Cell 2017; 31: 326-41. doi: 10.1016/j.ccell.2017.02.009 15. Lah TT, Duran Alonso MB, Van Noorden CF. Antiprotease therapy in cancer: hot or not? Expert Opin Biol Ther 2006; 6: 257-79. doi: 10.1517/14712598.6.3.257 16. Lah TT, Obermajer N, Duran Alonso MB, Kos J. Cysteine cathepsins and cystatins as cancer biomarkers. In: Edwards D, Hoyer-Hansen G, Blasi F, Sloane BF, editors. The cancer degradome: proteases and cancer biology. New York: Springer; 2008. p. 575-613. 17. Flannery T, McQuaid S, McGoohan C, McConnell RS, McGregor G, Mirakhur M, et al. Cathepsin S expression: an independent prognostic factor in glioblastoma tumours - a pilot study. Int J Cancer 2006; 119: 854-60. doi: 10.1002/ijc.21911 18. Verbovšek U, Motaln H, Rotter A, Atai NA, Gruden K, Van Noorden CJ, et al. Expression analysis of all protease genes reveals cathepsin K to be overexpressed in glioblastoma. PLoS One 2014; 9: e111819. doi: 10.1371/ journal.pone.0111819 19. Verbovšek U, Van Noorden CJ, Lah TT. Complexity of cancer protease biology: cathepsin K expression and function in cancer progression. Semin Cancer Biol 2015; 35: 71-84. doi: 10.1016/j.semcancer.2015.08.010 20. Novinec M, Lenarčič B. Cathepsin K: a unique collagenolytic cysteine peptidase. Biol Chem 2013; 394: 1163-79. doi: 10.1515/hsz-2013-0134 21. Staudt ND, Maurer A, Spring B, Kalbacher H, Aicher WK, Klein G. Processing of CXCL12 by different osteoblast-secreted cathepsins. Stem Cells Dev 2012; 21: 1924-35. doi: 10.1089/scd.2011.0307 22. Hira VV, Verbovšek U, Breznik B, Srdič M, Novinec M, Kakar H, et al. Cathepsin K cleavage of SDF-1a inhibits its chemotactic activity towards glioblastoma stem-like cells. Biochim Biophys Acta 2017; 1864: 594-603. doi: 10.1016/j.bbamcr.2016.12.021 23. Kos J, Vižin T, Fonovič UP, Pišlar A. Intracellular signaling by cathepsin X: molecular mechanisms and diagnostic and therapeutic opportunities in cancer. Semin Cancer Biol 2015; 31: 76-83. doi: 10.1016/j.semcancer.2014.05.001 Radiol Oncol 2018; 52(4): 433-442. 442 Breznik B et al. / Cathepsin K and X expression in glioblastoma 24. Kos J, Jevnikar Z, Obermajer N. The role of cathepsin X in cell signaling. Cell Adh Migr 2009; 3: 164-6. 25. Wendt W, Zhu XR, Lübbert H, Stichel CC. Differential expression of cathepsin X in aging and pathological central nervous system of mice. Exp Neurol 2007; 204: 525-40. doi: 10.1016/j.expneurol.2007.01.007 26. Nägler DK, Krüger S, Kellner A, Ziomek E, Menard R, Buhtz P, et al. Up-regulation of cathepsin X in prostate cancer and prostatic intraepithelial neoplasia. Prostate 2004; 60: 109-19. doi: 10.1002/pros.20046 27. Krueger S, Kalinski T, Hundertmark T, Wex T, Küster D, Peitz U, et al. Up-regulation of cathepsin X in Helicobacter pylori gastritis and gastric cancer. J Pathol 2005; 207: 32-42. doi: 10.1002/path.1820 28. Vizin T, Christensen IJ, Nielsen HJ, Kos J. Cathepsin X in serum from patients with colorectal cancer: relation to prognosis. Radiol Oncol 2012; 46: 207-12. doi: 10.2478/v10019-012-0040-0 29. Gopinath S, Malla R, Alapati K, Gorantla B, Gujrati M, Dinh DH, et al. Cathepsin B and uPAR regulate self-renewal of glioma-initiating cells through GLI-regulated Sox2 and Bmi1 expression. Carcinogenesis 2013; 34: 550-9. doi: 10.1093/carcin/bgs375 30. Alapati K, Kesanakurti D, Rao JS, Dasari VR. uPAR and cathepsin B-mediated compartmentalization of JNK regulates the migration of glioma-initiating cells. Stem Cell Res 2014; 12: 716-29. doi: 10.1016/j.scr.2014.02.008 31. Lathia JD, Mack SC, Mulkearns-Hubert EE, Valentim CL, Rich JN. Cancer stem cells in glioblastoma. Genes Dev 2015; 29: 1203-17. doi: 10.1101/ gad.261982.115 32. Godlewski J, Ferrer-Luna R, Rooj AK, Mineo M, Ricklefs F, Takeda YS, et al. MicroRNA Signatures and molecular subtypes of glioblastoma: the role of extracellular transfer. Stem Cell Reports 2017; 8: 1497-505. doi: 10.1016/j. stemcr.2017.04.024 33. Hira VV, Ploegmakers KJ, Grevers F, Verbovsek U, Silvestre-Roig C, Aronica E, et al. CD133+ and nestin+ glioma stem-like cells reside around CD31+ arterioles in niches that express SDF-1a, CXCR4, osteopontin and cathepsin K. J Histochem Cytochem 2015; 63: 481-93. doi: 10.1369/0022155415581689 34. Hira VVV, Aderetti DA, van Noorden CJF. Glioma stem cell niches in human glioblastoma are periarteriolar. J Histochem Cytochem 2018; 66: 349-58. doi: 10.1369/0022155417752676 35. Hira VVV, Wormer JR, Kakar H, Breznik B, van der Swaan B, Hulsbos R, et al. Periarteriolar glioblastoma stem cell niches express bone marrow hemat-opoietic stem cell niche proteins. J Histochem Cytochem 2018; 66: 155-73. doi: 10.1369/0022155417749174 36. GlioVis: Data visualization tools for brain tumor datasets. 2017. [cited 20 Sep 2017]. Available from: http://gliovis.bioinfo.cnio.es/. 37. Bowman RL, Wang Q, Carro A, Verhaak RG, Squatrito M. GlioVis data portal for visualization and analysis of brain tumor expression datasets. Neuro Oncol 2017; 19: 139-41. doi: 10.1093/neuonc/now247 38. Zavrsnik J, Butinar M, Trstenjak Prebanda M, Krajnc A, Vidmar R, Fonovic M, et al. Cystatin C deficiency suppresses tumor growth in a breast cancer model through decreased proliferation of tumor cells. Oncotarget 2017; 8: 73793-809. doi: 10.18632/oncotarget.17379 39. Chieco P, Jonker A, De Boer BA, Ruijter JM, Van Noorden CJ. Image cytometry: protocols for 2D and 3D quantification in microscopic images. Prog Histochem Cytochem 2013; 47: 211-333. doi: 10.1016/j.proghi.2012.09.001 40. Segerman A, Niklasson M, Haglund C, Bergström T, Jarvius M, Xie Y, et al. Clonal variation in drug and radiation response among glioma-initiating cells is linked to proneural-mesenchymal transition. Cell Rep 2016; 17: 29943009. doi: 10.1016/j.celrep.2016.11.056 41. Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, et al; Cancer Genome Atlas Research Network. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 2010; 17: 98-110. doi: 10.1016/j.ccr.2009.12.020 42. Zeppernick F, Ahmadi R, Campos B, Dictus C, Helmke BM, Becker N, et al. Stem cell marker CD133 affects clinical outcome in glioma patients. Clin Cancer Res 2008; 14: 123-9. doi: 10.1158/1078-0432.CCR-07-0932 43. Kenig S, Alonso MB, Mueller MM, Lah TT. Glioblastoma and endothelial cells cross-talk, mediated by SDF-1, enhances tumour invasion and endothe-lial proliferation by increasing expression of cathepsins B, S, and MMP-9. Cancer Lett 2010; 289: 53-61. doi: 10.1016/j.canlet.2009.07.014 44. Plaks V, Kong N, Werb Z. The cancer stem cell niche: how essential is the niche in regulating stemness of tumor cells? Cell Stem Cell 2015; 16: 225-38. doi: 10.1016/j.stem.2015.02.015 45. Saito K, Hirai T, Takeshima H, Kadota Y, Yamashita S, Ivanova A, Yokogami K. Genetic factors affecting intraoperative 5-aminolevulinic acid-induced fluorescence of diffuse gliomas. Radiol Oncol 2017; 51:142-50. doi: 10.1515/ raon-2017-0019 46. Strojnik T, Kos J, Zidanik B, Golouh R, Lah T. Cathepsin B immunohistochemi-cal staining in tumor and endothelial cells is a new prognostic factor for survival in patients with brain tumors. Clin Cancer Res 1999; 5: 559-67. 47. Strojnik T, Kavalar R, Trinkaus M, Lah TT. Cathepsin L in glioma progression: comparison with cathepsin B. Cancer Detect Prev 2005; 29: 448-55. doi: 10.1016/j.cdp.2005.07.006 48. Dauth S, Schmidt MM, Rehders M, Dietz F, Kelm S, Dringen R, et al. Characterisation and metabolism of astroglia-rich primary cultures from cathepsin K-deficient mice. Biol Chem 2012; 393: 959-70. doi: 10.1515/ hsz-2012-0145 49. Dauth S, Sîrbulescu RF, Jordans S, Rehders M, Avena L, Oswald J, et al. Cathepsin K deficiency in mice induces structural and metabolic changes in the central nervous system that are associated with learning and memory deficits. BMC Neurosci 2011; 12: 74. doi: 10.1186/1471-2202-12-74 50. Wang J, Chen L, Li Y, Guan XY. Overexpression of cathepsin Z contributes to tumor metastasis by inducing epithelial-mesenchymal transition in hepatocellular carcinoma. PLoS One 2011; 6: e24967. doi: 10.1371/journal. pone.0024967 51. Sevenich L, Schurigt U, Sachse K, Gajda M, Werner F, Müller S, et al. Synergistic antitumor effects of combined cathepsin B and cathepsin Z deficiencies on breast cancer progression and metastasis in mice. Proc Natl Acad Sci U S A 2010; 107: 2497-502. doi: 10.1073/pnas.0907240107 52. Mitrovic A, Pecar Fonovic U, Kos J. Cysteine cathepsins B and X promote epithelial-mesenchymal transition of tumor cells. Eur J Cell Biol 2017; 96: 622-31. doi: 10.1016/j.ejcb.2017.04.003 53. Akkari L, Gocheva V, Kester JC, Hunter KE, Quick ML, Sevenich L, et al. Distinct functions of macrophage-derived and cancer cell-derived cathepsin Z combine to promote tumor malignancy via interactions with the extracellular matrix. Genes Dev 2014; 28: 2134-50. doi: 10.1101/gad.249599.114 54. Euskirchen P, Radke J, Schmidt MS, Schulze Heuling E, Kadikowski E, Maricos M, et al. Cellular heterogeneity contributes to subtype-specific expression of ZEB1 in human glioblastoma. PLoS One 2017; 12: e0185376. doi: 10.1371/journal.pone.0185376 55. Calabrese C, Poppleton H, Kocak M, Hogg TL, Fuller C, Hamner B, et al. A perivascular niche for brain tumor stem cells. Cancer Cell 2007; 11: 69-82. doi: 10.1016/j.ccr.2006.11.020 56. Zhu TS, Costello MA, Talsma CE, Flack CG, Crowley JG, Hamm LL, et al. Endothelial cells create a stem cell niche in glioblastoma by providing NOTCH ligands that nurture self-renewal of cancer stem-like cells. Cancer Res 2011; 71: 6061-72. doi: 10.1158/0008-5472.CAN-10-4269 57. Goffart N, Lombard A, Lallemand F, Kroonen J, Nassen J, Di Valentin E, et al. CXCL12 mediates glioblastoma resistance to radiotherapy in the subven-tricular zone. Neuro Oncol 2017; 19: 66-77. doi: 10.1093/neuonc/now136 58. Ardebili SY, Zajc I, Gole B, Campos B, Herold-Mende C, Drmota S, et al. CD133/prominin1 is prognostic for GBM patient's survival, but inversely correlated with cysteine cathepsins' expression in glioblastoma derived spheroids. Radiol Oncol 2011; 45: 102-15. doi: 10.2478/v10019-011-0015-6 59. Kollet O, Dar A, Shivtiel S, Kalinkovich A, Lapid K, Sztainberg Y, et al. Osteoclasts degrade endosteal components and promote mobilization of hematopoietic progenitor cells. Nat Med 2006 ;12: 657-64. doi: 10.1038/ nm1417 60. MEROPS database: The peptidase database. [cited 15 Dec 2017]. Available at https://www.ebi.ac.uk/merops/index.shtml. Radiol Oncol 2018; 52(4): 433-442.