miR‐218 affects the ECM composition and cell biomechanical properties of glioblastoma cells

Abstract Glioblastoma (GBM) is the most common malignant brain tumour. GBM cells have the ability to infiltrate into the surrounding brain tissue, which results in a significant decrease in the patient’s survival rate. Infiltration is a consequence of the low adhesion and high migration of the tumour cells, two features being associated with the highly remodelled extracellular matrix (ECM). In this study, we report that ECM composition is partially regulated at the post‐transcriptional level by miRNA. Particularly, we show that miR‐218, a well‐known miRNA suppressor, is involved in the direct regulation of ECM components, tenascin‐C (TN‐C) and syndecan‐2 (SDC‐2). We demonstrated that the overexpression of miR‐218 reduces the mRNA and protein expression levels of TN‐C and SDC‐2, and subsequently influences biomechanical properties of GBM cells. Atomic force microscopy (AFM) and real‐time migration analysis revealed that miR‐218 overexpression impairs the migration potential and enhances the adhesive properties of cells. AFM analysis followed by F‐actin staining demonstrated that the expression level of miR‐218 has an impact on cell stiffness and cytoskeletal reorganization. Global gene expression analysis showed deregulation of a number of genes involved in tumour cell motility and adhesion or ECM remodelling upon miR‐218 treatment, suggesting further indirect interactions between the cells and ECM. The results demonstrated a direct impact of miR‐218 reduction in GBM tumours on the qualitative ECM content, leading to changes in the rigidity of the ECM and GBM cells being conducive to increased invasiveness of GBM.


| INTRODUC TI ON
Glioblastoma (GBM) is the most malignant astrocytic brain tumour.
Despite treatment with advanced therapies, including aggressive surgical intervention, radiotherapy, and systemic chemotherapy, as well as significant advances in the field of oncology, the average survival time for GBM patients is approximately 15 months, with a 5-year survival rate of only 5%. 1,2 The main factor contributing to this poor prognosis is the ability of GBM cells to infiltrate adjacent tissues, resulting in a high rate of tumour recurrence. 3 These notable migration and invasion abilities could be explained by alterations occurring in the structure of cancer cells and their surroundings, defined by mechanobiology. 4 To promote the invasiveness, cancer cells modify not only themselves but also their environment, namely the extracellular matrix (ECM). It consists of over 300 different proteins, including proteoglycans and glycoproteins. 5 Neoplastic tissues are characterized by the phenomenon of desmoplasia, manifested by the intense formation of a dense ECM consisting of increased levels of total fibrillar collagen, fibronectin, proteoglycans and tenascin-C (TN-C). 6 The capability to synthesize specific and cancer-related ECM components has been shown to be relevant for the high invasiveness of tumour cells. The changed protein profile within ECM increases the stiffness of cancerous tissue, 7,8 which may lead to enhanced cell-ECM adhesion through the involvement of local adhesion proteins. The general trend observed for many types of cells indicates that cell spread and adhesion are improved on harder matrices. 9,10 The effect of the environment on the cells is explained by the mechanotransduction mechanism, in which mechanical and cell-specific signals are actively detected by cells and converted into intracellular biochemical signals. In this manner, the ECM can affect cancer cell behaviour, including invasion and metastasis. 11,12 Therefore, cancer should be considered as a disease with alterations in both cells and their microenvironment, including also the biochemical and biophysical properties of the ECM. Not only proteins suspended in the ECM have an impact on the invasiveness of the tumour, but also transmembrane proteins. The syndecans are a four-member family of evolutionarily conserved small type I transmembrane proteoglycans implicated in the formation of specialized membrane domains, cell adhesion, cytoskeletal organization, migration and wound healing. They have been also related to the pathological conditions, including inflammation and cancer. [13][14][15] For instance, elevated expression of syndecan-2 (SDC-2) has been correlated with increased invasiveness in various types of cancers, including fibrosarcoma, 16 melanoma, 17 colon, 18 pancreatic 19 and colorectal 20 cancers, while TN-C is overexpressed in brain tumours, 21 breast, 22 lung 23 and colorectal 24 cancers.
In the recent 20 years, microRNAs (miRNAs) have emerged as key regulators of gene expression at the post-transcriptional level.
miRNAs are a large family of endogenous, evolutionarily conserved, non-coding RNAs that are ~22 nucleotides long, and they have been implicated in the regulation of nearly every biological process. 25 Deregulated miRNA expression has been shown to play a role in the pathogenesis of a growing list of human diseases, including cancer and cardiovascular, neurodegenerative and autoimmune disorders. [26][27][28][29] For example, in GBM, it has been already demonstrated that the downregulation of miR-218 affects cell proliferation, epithelial-to-mesenchymal transition, 30 metabolism of cancer cells 31 and cancer stem cell properties. 32 How miRNAs are involved in the regulation of ECM composition and the mechanobiology of cancer cells in GBM tumours is largely unknown. In principle, miRNAs can exert their control over the ECM either directly by targeting mRNAs encoding ECM proteins or indirectly by modulating the expression of genes involved in the synthesis or degradation of ECM molecules.
Here, we have evidenced that miR-218, one of the highly downregulated miRNAs in GBM cells, is involved in the direct regulation of TN-C and SDC-2, two highly overrepresented proteins in GBM and ECM components. Both SDC-2 and TN-C have been previously demonstrated to increase tumour cell migration and invasiveness.
In the course of the study, we attempted to validate how miR-218 interaction with its ECM targets affects globally a microenviron- Collectively, our results indicate that miR-218 is a potent tumour suppressor in glioma with a substantial impact on the ECM composition and biomechanical properties of GBM. penicillin-streptomycin antibiotic (Sigma-Aldrich) and incubated at 37°C and 5% CO 2 in a humidified atmosphere in an incubator.

| Transfection
The cells were transfected with mirVana™ hsa-miR-218-5p mimic (Invitrogen) in a final concentration of 10 nM and 50 nM at 70%-80% confluency. Lipofectamine™ 2000 (Invitrogen) was used as a transfection agent according to the manufacturer's protocols. A non-specific scrambled siRNA (Sigma-Aldrich) was used as a control in all transfection experiments. The cells were processed after 24 h for the quantification of transcript levels using qPCR, Western blot, cellular assays or AFM analysis.
The intensity of individual bands was analysed quantitively by Multi Gauge ver. 2.0 (Fujifilm). The relative ratio of protein-level expression was determined based on the densitometric measurements of band intensities in relation to the control sample.

| Real-time migration
Real-time cell migration monitoring was performed in the xCELLi-gence® system using the RTCA DP apparatus (ACEA Biosciences).
The experiment was carried out on 16

| Real-time proliferation
The use of the xCELLigence® system enabled the observation of real-time cell proliferation. In that experiment were used the E-

| Real-time adhesion
The xCELLigence® system together with the E-Plates PET (ACEA Biosciences) was used. In each well of them, four rows of microelectrode sensors are removed, creating a window for cell visualization.
Plates were covered with poly-L-lysine (Sigma-Aldrich), incubated for one hour in 37°C and rinsed with phosphate-buffered saline (PBS, VWR Life Science). Additionally, some wells were overlaid with 1% BSA for 20 min and acted as a negative control. An unsupplemented medium was analysed as a background. Then, 24 h earlier transfected cells were seeded 10,000 cells per well in serum-free medium. Measurements took place every three minutes for four hours.

| Single-cell force spectroscopy (SCFS)
Cell deformability and adhesiveness were determined from the AFM measurements carried out in single-cell force spectroscopy (SCFS)

| Gene expression analysis based on the Gliovis database
TN-C and SDC-2 gene expression analysis for large groups of patients was performed using the Gliovis online tool (gliov is.bioin fo.cnio.es). In order to comprehensively present the expression of these genes in large group of tissues, databases from two independent projects (TCGA and Rembrandt) were used. Data set HG-U133A contained information from 10 non-tumour samples and 528 samples described as GBM. The Agilent-4502A data set contained the same number of healthy samples and 489 samples of glioblastoma.
The Rembrandt database, on the contrary, has a set of 28 healthy samples, 225 described as non-tumour glioma and 219 GBM samples. Only the "non-tumour" and "GBM" data sets were used from the Rembrandt database.

| Statistical analysis
The results are presented as a mean value ± standard deviation (SD).

| In glioblastoma, the expression level of miR-218 correlates inversely with the expression levels of the ECM components TN-C and SDC-2
Our previous study revealed that there were 97 miRNAs differentially expressed in glioblastoma compared with those in the healthy brain. 36 Forty-one of these miRNAs showed a reduced expression level in malignant gliomas. Among these miRNAs, we found miR-218 to be significantly downregulated in brain tumour tissues. We further confirmed the expression level of miR-218 in primary and recurrent GBM via qRT-PCR analysis ( Figure 1A). The levels of miR-218 expression in primary tumour tissue and recurrent GBM tissue were 56% and 69% lower, respectively, than those in healthy brain tissue.
Given the profound downregulation of miR-218 in GBM, we sought to investigate its putative targets. To identify the binding sites in the 3′UTRs of genes that can be potentially regulated by selected miRNAs, we used prediction software such as ENCORI, miRDB, PicTar and TargetScan. Interestingly, among the predicted targets, we found several genes encoding ECM proteins, such as

| TN-C and SDC-2 are direct targets of miR-218
Highly ranked binding targets of miR-218 were subjected to further analysis. We focused specifically on TN-C and SDC-2 and investigated their gene expression levels via qRT-PCR. The tenascin-C level was significantly increased in all examined tumour samples-that is eightfold higher in primary tumour tissue and 21-fold higher in recurrent tumour tissue than in healthy All algorithms used for miR-218 target prediction showed one binding site within the 3′UTR of both the TN-C and SDC-2 mRNAs.
We employed a set of reporter constructs in a luciferase assay to experimentally verify the predicted binding of miR-218 to its target sites within the 3′UTRs of TN-C and SDC-2. The following constructs were tested in parallel: wild-type reporters (WT) containing a single native binding site for either miR-218, constructs with mutations (MUT) disrupting the 5′ seed site (negative controls) and con-

| miR-218 affects the ECM composition
Given the above results, we have evaluated the miR-218 overexpression on the ECM composition. To test this hypothesis, we used a We further hypothesized that changes in the ECM composition due to miR-218 overexpression also affect the mechanobiological properties of cancer cells.

| Impaired cell migration after miR-218 treatment
To explore the impact of miR-218 on cell migration, we compared the migration rate of miR-218-transfected U-118 MG cells with that of non-treated (negative control) cells ( Figure 4A). The mathematical interpretation of the impedance (CI value) for each experimen- A [methyl-3H]-thymidine incorporation assay was performed to complement the proliferation analysis with the xCELLigence system.
In this study, the degree of incorporation of radioactively labelled thymidine was evaluated and translated into the replication potential of cells. In addition to the standard trials used, we analysed the effect of 3 μM camptothecin, which has a confirmed pro-apoptotic effect, 37 as a positive control in the experiment. The incorporation rate in CPT-treated cells was 69% compared with that in control cells. miR-218 mimic transfection increased the incorporation of tritiated thymidine by 57% at 10 nM and 49% at 50 nM compared with that in control cells ( Figure 4F).

| miR-218 enhances glioma cell adhesion
To explore the impact of miR-218 on U118-MG cells, the cell surface properties were quantified using AFM in SCFS mode ( Figure 5D).

| Overexpression of miR-218 impacts cell stiffness
Most surface receptors are linked not only to ECM proteins but also to actin filaments forming the actin cortex. 38 Thus, in our next step, we verified whether changes in GBM cell adhesive properties contribute to the overall mechanical properties of these cells. Cell stiffness was measured for cells compressed between the surface and a tipless cantilever; therefore, it was calculated as the slope of the approach part of the recorded force curves and expressed in N/m ( Figure 6D). ( Figure 6D). As the cell stiffness measured using AFM is related to the organization of the actin cytoskeleton, it was further visualized using fluorescently labelled F-actin to verify the effect ( Figure 6E). The observed changes confirm then the hypothesis that cell stiffness is related to the cytoskeleton.

| DISCUSS ION
The malignancy of glioblastoma depends on its ability to infiltrate adjacent tissues and to create secondary lesions. 39 The aggressive growth of glioma tumours and difficulties in developing an effective treatment scheme have led to intense integration of medical and molecular biological research. Remodelling of the ECM and miRNA deregulation are known processes contributing to GBM cell invasion and brain infiltration. [40][41][42] In this study, we show that miR-218 can The positive correlation between SLIT2 and miR-218 expression has been shown, what indicates that these two molecules are transcribed together. 45 The SLIT2 downregulation in GBM in consequence leads to the further decreased expression of miR-218. 46 Moreover, the expression level of miR-218 in GBM might be invoked by the feedback mechanism. The decreased expression of miR-218 can directly increase the expression of effector molecules such as RSK2, 6SK1 and PDGFRα, maintaining then the activity of the RTK pathway at a high level. RTK-conducted signals stimulate the expression of the STAT3 gene, whose product together with BCLAF1 binds directly to the miR-218 locus, thereby suppressing its expression. 47 Our previous finding confirmed then the decreased expression of miR-218 both in primary and in recurrent tumours by 50% and 70%, respectively. Decreased miR-218-5p expression levels have also been reported in other types of human cancer, such as medulloblastoma, thyroid cancer and cervical cancer. [48][49][50] We confirmed that the predicted miR-218 targets, the ECM components TN-C and SDC-2, are directly regulated by miR-218. We used a dual-luciferase assay and miR-218 mimic to verify these functional interactions. The effects were detectable at both the mRNA and protein levels for both TN-C and SDC-2. Proteins derived from these transcripts are potentially key factors in the ECM of cancer cells. 15,21 The presence of TN-C in cancer tissues was initially considered as a characteristic feature of only gliomas, 51 with its expression increasing in proportion to the degree of brain tumour malignancy. 52 Its presence was found to increase the proliferation and invasiveness of cancer cells and to take part in the process of angiogenesis. 53 The The obtained data revealed the decrease in the rate of cell migration upon the overexpression of miR-218, but at the same time also an increase in their proliferation potential (Figure 4). Our observations seem to be consistent with the "go-or-grow" hypothesis, according to which the division of neoplastic cells and their movement are two temporally exclusive events. 63 The "go-or-grow" decision is strictly regulated and modulated by changes in the tumour microenvironment, which allows cells to "go" towards more favourable conditions to proliferate at the distant site or to "grow" and to stay at the site of origin, if their current environment provides the proper conditions for tumour growth. Changes in miRNA expression, followed by the ECM remodelling, can modulate the "go-or-grow" decision. As it has already been shown previously, the considerable overexpression of miR-9 in glioma cells inhibits proliferation but concurrently promotes migration. 64 Evidence indicates that mechanical properties and deformability can also be used as biomarkers to distinguish between healthy and cancer cells. The deformability of a whole cell, which depends on the properties of the cytoplasm, the cytoskeleton and the nucleus, can be defined in terms of the response of the cell to an applied stress. One of the techniques that enables the measurement of biophysical properties of cells, such as adhesion and stiffness, is AFM. 65 We evaluated the mechanobiological properties of GBM cells, including adhesion and stiffness, upon miR-218 mimic treatment. We obtained real-time measurements in cell culture can thus impact the adhesion of that cell. 68,69 It has already been shown that tumours can become stiffer than normal tissues due to increased Rho-dependent cytoskeletal pressure, generating excessive growth, focal adhesions, adjacent joint division and tissue disruption. 70 Stiffness also directly depends on the malignancy of the tumour. It is known that invasive GBM tumours produce stiffnesspromoting factors such as collagen, fibronectin and laminins, which may suggest that the production of these proteins is disrupted after miR-218 overexpression. 71 An increase in stiffness has also been observed in many different types of cancer cells, such as breast cancer, melanoma, prostate cancer and cervical cancer cells. An important aspect of cell stiffness is the ratio of cancer to normal cells. While cancer cells are less stiff than normal cells, 72 the same pattern of stiffness is also observed in malignant versus non-malignant tissues in breast cancer, 73 bladder cancer 74 and prostate cancer. 75 In our research, glioblastoma cells with miR-218 overexpression were approximately 30% stiffer than non-treated cells. Increased stiffness in brain tissues can be correlated with diseases such as brain abscess or with cytoskeletal maturation in brain cells. 76 The correlation of cytoskeletal maturation with an increase in cell stiffness has been observed for astrocytes, in which the AFM-measured stiffness may increase sevenfold in a 5- week observation period during development. 77 In miR-218-treated GBM cells, the actin cytoskeleton was slightly rearranged, which could explain the increase in cell stiffness.
The minor discrepancy in the relation between cell surface ad- Collectively, these results demonstrated that miRNA-218 strongly affects the expression of genes encoding cell surface receptors responsible for the adhesive properties of cells.
We also found that miR-218-treated cells are more rigid than non-treated cells, which presumably might prevent them from undergoing extravasation and intravasation during migration and invasion events. We thus hypothesized that miR-218 overexpression can support the maintenance of the non-invasive cell phenotype, which is correlated with differences in mechanical properties. The observation is more important when one realizes the importance of ECM rigidity in the perivascular space. It has already been shown that this part of the brain tissue is more rigid in GBM, thus promoting glioma cell migration. 62 As we have shown, miR-218-5p deregulation is involved in GBM growth and migration potential.  93,94 The STAT3 level is correlated with GBM malignancy, indicating its participation in increasing the migration potential of cancer cells. 95 Additionally, regarding EGF, its impact on the migratory nature of GBM cells is known. 96 CTNN and the Arp2/3 complex are known for regulating lamellipodia formation, and a decrease in CTNN expression can suppress GBM migration mechanisms. 97,98 Because we showed a decrease in the migration capacity of glioblastoma cells under treatment with miR-218 in our studies, we could conclude then that these changes might be the result of the impact of miR-218 on CDC42, STAT3, EGF and CTTN.
The observed increase in GBM cell adhesion may also be associated with a decrease in ACTN1 expression. It has been shown that after downregulation of ACTN1, GBM cells show poor spread but increased focal adhesion. 99 The changes in the cytoskeleton that we

ACK N OWLED G EM ENTS
We thank the Laboratory of Subcellular Structures Analysis of the Institute of Bioorganic Chemistry, Polish Academy of Sciences in Poznan, for facilitating the cell culture and maintaining the microscopic analysis.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data available on request from the authors.

I N S TITUTI O N A L R E V I E W B OA R D S TATE M E NT
Tissue samples used in this study were collected based on the ap-