Integrin α 5 β 1 plays a critical role in resistance to temozolomide by interfering with the p 53 pathway in high grade glioma

Integrins play a role in the resistance of advanced cancers to radiotherapy and chemotherapy. In this study, we show that high expression of the α5 integrin subunit compromises temozolomide-induced tumor suppressor p53 activity in human glioblastoma cells. We found that depletion of the α5 integrin subunit increased p53 activity and temozolomide sensitivity. However, when cells were treated with the p53 activator nutlin-3a, the protective effect of α5 integrin on p53 activation and cell survival was lost. In a functional p53 background, nutlin-3a downregulated the α5 integrin subunit, thereby increasing the cytotoxic effect of temozolomide. Clinically, α5β1 integrin expression was associated with a more aggressive phenotype in brain tumors, and high α5 integrin gene expression was associated with decreased survival of patients with high-grade glioma. Taken together, our findings indicate that negative cross-talk between α5β1 integrin and p53 supports glioma resistance to temozolomide, providing preclinical proof-of-concept that α5β1 integrin represents a therapeutic target for high-grade brain tumors. Direct activation of p53 may remain a therapeutic option in the subset of patients with high-grade gliomas that express both functional p53 and a high level of α5β1 integrin.


INTRODUCTION
Glioblastoma multiforme (GBM) are the most aggressive brain tumors and remain a challenge for oncologists.New therapies are urgently needed.Gene expression profiling of high-grade glioma revealed that genes of extracellular matrix components and their regulators are often affected in the patients.Fibronectin is overexpressed in glioblastoma versus normal brain (1), and belongs to the cluster of genes associated with a more malignant phenotype (2,3).It has recently been shown that fibronectin knockdown delays tumor growth in a mouse glioma model (4).The α5β1 integrin is a fibronectin receptor which was recently shown to have an important role in tumor progression (5), metastasis (6) and/or resistance to therapies (7) in lung, ovarian and breast cancer, respectively.Few works addressed directly the issue of α5β1 integrin in glioma.Through the use of non-peptidic α5β1 integrin antagonists and GBM cell lines, we previously demonstrated that α5β1 integrin may be a therapeutic target for these tumors (8,9) and that concomitant addition of α5β1 antagonists sensitizes p53 wild-type (p53 wt) glioma cells to chemotherapeutic drugs (10).The presence of p53 mutations in high-grade glioma varied across GBM subtypes with 0, 21, 32 and 54 % in classical, neural, mesenchymal and proneural subtypes, respectively (11).There is increasing evidence that gliomas harboring a p53 wt resist to therapies through inhibitory pathways upstream of p53.Nutlin-3 belongs to the family of small-molecule inhibitors of the mdm2-p53 interaction (12).Nutlin-3 has been shown, alone or in combination with chemotherapeutic agents, to increase the degree of apoptosis in hematological malignancies (13).Recent studies extended its therapeutic window for use in solid tumors (14,15).
The aim of this study was to investigate the role of α5β1 integrin in glioma resistance to Temozolomide (TMZ) chemotherapy using in vitro and in vivo models.We found Research.
on January 22, 2018.© 2012 American Association for Cancer cancerres.aacrjournals.orgDownloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
Author Manuscript Published OnlineFirst on May 16, 2012; DOI: 10.1158/0008-5472.that a high expression of α5 subunit inhibited the TMZ-induced p53 pathway and that reactivation of p53 by nutlin-3a restores the sensitivity to TMZ by decreasing the expression of the α5β1 integrin.Finally we found that high α5 integrin gene expression is associated with a more aggressive phenotype in brain tumors and a decrease in survival of patients.Our results provide a clinical rationale for including α5β1 integrin-targeted therapy in a subpopulation of glioma patients.

Reagents
TMZ was a kind gift from Schering-Plough.Nutlin-3a, the active enantiomer of nutlin, was from Cayman.TMZ was prepared before use at 10mM in 50/50 ethanol/H 2 O.
Other drugs were prepared as stock solutions in ethanol at 10mM and were kept at -20°C until use.

Cell culture and transfection
The U87MG cells (p53 wt) was from ATCC; the U373 cells (p53 mutated) from ECACC (Sigma) and not authenticated in the laboratory.The LN18 (p53 mutated) and LNZ308 (p53 KO) cells were kindly provided by Pr.M. Hegi (Switzerland).Cells were cultured as described elsewhere (10).The identity of cell lines was regularly checked by morphologic criteria and importantly p53 status was routinely checked either by the yeast functional assay (16), western blot quantification of p53 stability and phosphorylation and by qPCR quantification of p53 target genes after treatment with ellipticine.Cells were stably transfected to overexpress (by transfecting a pcDNA3.1 plasmid containing the human α5 integrin gene, provided by Dr. Ruoshlati, La Jolla) or to repress (by transfecting a pSM2 plasmid coding for a shRNA targeting the α5 mRNA, OpenBiosystems) the α5 integrin subunit by using jetPRIME™ (Polyplus-transfection) according to the manufacturer's instructions.The vector for the p53-wt transfection was a kind gift from Dr. C. Blattner (Germany).Cells were transfected with specific siRNA for human p53, the α5 integrin subunit or nontargeting siRNA (Thermo Scientific Dharmacon) with jetPRIME™ (Polyplustransfection) according to the manufacturer's instructions.Western blotting was carried out as previously described (10).Antibodies used were against α5 integrin Ab1928 (Millipore) or H104 (Santa-Cruz), β1 integrin Ab1952, GAPDH (Millipore,), p53 (BD Biosciences), or p53 pser15 (Cell Signaling).

Flow cytometry
After detachment with EDTA, cells were incubated for 30 min at 4°C under agitation in the presence of primary antibodies: anti-α5 integrin antibody IIA1 (BD Biosciences) and anti-β1 integrin antibodies (TS2/16 from Santa Cruz; 9EG7 and mab13 from BD Biosciences).After washing, cells were incubated for 30 min with secondary antibody (Alexa488-conjugated goat anti-mouse or rat, Jacksonimmuno Research).After washing, cells were analyzed using a FACS Calibur flow cytometer (Becton-Dickinson), and the mean fluorescence intensity characterizing surface expression of integrins was measured using the Cell Quest software.

Immunofluorescence
A total of 20 000 cells were seeded onto IBIDI µdishes coated with 10 µg/ml of poly-L-lysine.Cells were treated with nutlin-3a (10 µM) or with solvent during 24 hours before fixation with 4% PFA (10 min at room temperature) and then processed for α5 immunodetection (IIA1 antibody; 1/300).Confocal images were taken with a confocal microscope (Biorad 1024) equipped with a water immersion 60x objective.Images were collected using the Laser-Sharp 2000 software.

Human biopsies
This study was conducted on 115 adult brain biopsies, 95 brain tumors (22 grade II, 38 grade III and 35 grade IV) and 20 non-tumoral brain tissues collected retrospectively from archival material stored at the Centre de Ressources Biologiques et Tumorothèque (Hopitaux Universitaires de Strasbourg).The patient characteristics have been described elsewhere (17) pathologist to specify the tumor grade and the percentage of tumor cells.Only samples with at least 50% of tumoral cells (more than 50% of samples were > 70% tumoral cells) have been included in the study.Control tissues were obtained from epileptic surgery.The study was conducted in accordance with the Declaration of Helsinki.Realtime quantitative PCR was performed as described previously (17).The threshold cycle (Ct) values for each gene were normalized to expression level of cyclophilin used as the housekeeping gene.Values were normalized relatively to the value obtained for one non tumoral control brain tissue which was included in each qPCR run.Immunological analysis of α5 protein expression was done as shown previously (17).

Human brain tumor data sets
Glioma gene expression datasets from two other cohorts were downloaded from the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/GEO,accession numbers GSE4271 and GSE4412).Micro array raw data were processed using R (version 2.10.1)(http://cran.r-project.org/),implemented with the BioConductor package (http://www.bioconductor.org).Estimates of the survival curves were computed using the Kaplan-Meier method.Univariate survival comparisons between the patients, according to low or high α5-integrin expression levels, were performed using a logrank test.
Human brain tumor xenografts TCG4, TCG9 and TCG17 glioma xenograft models were obtained as previously described (18).Subcutaneous tumor growth was followed by measuring, three times per week, two perpendicular diameters with a caliper.Treatments began when tumors reached a volume of approximately 250 ± 50 mm 3 .TMZ was administered p.o. at the dose of 50 mg/kg/day for 5 days.Mice were sacrificed when the tumor volumes reached 4 times their initial volume (V0).For each mouse, the time between the treatment onset and the animal sacrifice was defined as the "survival time".TP53 status of each xenograft was determined by the yeast functional assay (16).

Statistical analysis
Data are represented as the mean ± SEM, and n is the number of independent experiments.Statistical analyses were conducted using Student's t test or the Mann-Whitney test with the GraphPad Prism program.p < 0.05 was considered significant.

α5β1 integrin impedes TMZ-induced p53wt activity.
We compared the effect of TMZ in U87MG cells depleted in (shRNA α5 ) or overexpressing (pcDNA α5 ) α5 integrin versus control cells (shRNA ns and pcDNA ctrl respectively).Because p53 is largely involved in chemotherapeutic drug effects and we showed previously that α5β1 integrin antagonists modulate the p53 pathway (10), we focused on TMZ-induced p53 activation.TMZ caused an increase in p53 protein in all cell lines but not significantly in pcDNA α5 cells (when normalized to GAPDH -Figure 1a).Interestingly, a significant increase in p53 protein was already observed in untreated shRNA α5 cells versus shRNA ctrl cells (Figure 1a).After TMZ treatment, an increase in p53 pser15 was detectable in pcDNA ctrl and shRNA ns cells, which was significantly more pronounced in shRNA α5 (Figure 1a).In contrast, in pcDNA α5 cells, significantly less p53 pser15 was measured after TMZ treatment.Transcriptional activity of p53 was higher in shRNA α5 cells and lower in pcDNA α5 cells as compared to TMZtreated control cells (Figure 1b).Taken together, these results indicate that α5β1 integrin modulates p53 activity and that high expression of this integrin inhibits TMZinduced p53 stimulation.Modulation of p53 activity was related to cell survival as pcDNA α5 cells are significantly more resistant at high TMZ concentration while shRNA α5 cells appear more sensitive than their control counterparts (Figure 1c).α5 integrin overexpression did not modulate p53 nor clonogenic survival in U373 and LN18 cells expressing a p53 mutant (Supplementary Figure S1).In addition, repression of α5 integrin in p53-deficient LNZ308 cells did not sensitize cells to TMZ (supplementary Figure S1).From these data, we concluded that α5β1 integrininduced TMZ resistance requires a functional p53.As a first approach to confirm the role of the α5β1 integrin in TMZ chemoresistance in vivo, we used subcutaneous xenografted human brain tumors in nude mice.We selected three xenografts that exhibited a wild-type p53 and different levels of the α5 subunit.Kaplan-Meier analysis of the mouse survival suggest a relationship between α5 integrin level and resistance to TMZ (Figure 1d), providing some evidence for a role of the α5 integrin in the chemoresistance of p53-wt-expressing tumors in vivo.

Activation of p53 by Nutlin-3a overrides the α5 integrin effects.
We next investigated whether high α5 also impacts on p53 activation by a nongenotoxic p53 activator in glioma cells.U87MG cells were treated with nutlin-3a.In contrast to the effects of TMZ, nutlin-3a stabilized p53, markedly increased the p53 pser15 and the transactivation of p53 target genes in pcDNA ctrl and in pcDNA α5 cells (Figure 2a and b).Addition of TMZ to nutlin-3a does not further increase these effects (Figure 2a).In clonogenic assays, the α5β1 integrin did not efficiently protect the cells from death when p53 was activated by 10 µM nutlin-3a (Figure 2c).Survival of LNZ308 cells (p53 KO) or U373 cells (p53 mutant) were less affected after treatment with nutlin-3a (Figure 2c).

Activation of p53 by Nutlin-3a markedly decreases the α5 expression level in glioma cells.
U87MG cells treated with nutlin-3a rounded up and detached from the wells.This effect was lost when p53 expression was inhibited with specific siRNA (Figure 3a, left).In contrast, LNZ308 cells did not exhibit any morphological alterations after a 10 µM nutlin-3a treatment unless p53 was re-expressed in the cells (Figure 3a, right).
Interestingly, cell treatment with nutlin-3a decreased the expression of the α5 integrin at the protein level in U87MG-pcDNA ctrl and U87MG-pcDNA α5 cells, an effect not observed after treatment with TMZ (Figure 3b).expression after nutlin-3a treatment was confirmed by specific immunostaining of the α5 subunit in U87MG-pcDNA ctrl and U87MG-pcDNA α5 cells and by flow cytometry analysis of the α5 subunit at the cell membrane (Figure 3b).A significant decrease in the α5 mRNA level was measured in U87MG pcDNA ctrl but not in pcDNA α5 cells, suggesting that nutlin-3a affected the α5 subunit at translational and posttranslational levels (Figure 3c).Nutlin-3a also decreased β1 at the protein and mRNA level in the U87MG cells, suggesting that both subunits of the α5β1 integrin are processed similarly after nutlin-3a treatment (Supplementary Figure S2).However no effect on the β1 subunit expressed at the cell membrane could be detected after Nutlin-3a treatment (Supplementary Figure S2 and supplementary Table 1).
Nutlin-3a did not affect the endogenous α5 protein in p53 knockout LNZ308 cells unless p53 was re-expressed (Figure 3d).Nutlin-3a had no effect on α5 expression in U373 cells (Supplementary Figure S3).Altogether, these data suggest that nutlin-3a requires a functional p53 to decrease α5 expression, which in turn make the cells susceptible to the nutlin-3a-induced cell death.

High α5β1 integrin expression is associated with worse clinical outcome in high-grade glioma.
To our knowledge, no studies have associated α5β1 integrin with clinical outcome in patients with glioma.To investigate first whether integrin expression is associated with the grade of brain tumors, gene expression of the α5 and β1 subunits were examined by qPCR in 95 human brain tumors of different grades and compared to 20 non-tumor brain samples.The data revealed that α5 subunit gene expression was increased with increasing tumor grade although the β1 subunit was equally overexpressed in the three tumoral grades compared to control tissue (Figure 4a).Data were confirmed at the protein level (Figure 4b).Because the α5 subunit only dimerizes with β1, the data point towards a particular role for α5β1 integrin in glioma.
We analyzed the clinical data of grade III and grade IV patients.Log-rank analysis of the Kaplan-Meier survival curves demonstrated a significant survival advantage for patients with low α5-expressing glioma compared to high α5-expressing glioma.
These results were validated in two independent public datasets (2,19).Considering all three cohorts together, the group of high α5-expressing tumors included 39% of grade 3 and 81% of grade 4 tumors (Figure 4c and supplementary Table 2).Finally, we evaluated the relationship between α5 integrin level and the status of p53 in 56 human biopsies (grade III and IV) and in 17 human tumor xenografts.A clear tendency towards a higher level of α5 in p53 wt versus p53 mutant tumors was found in biopsies and in xenografts (Figure 4d and supplementary Table 3).

DISCUSSION
The data summarized here document the impact of the α5β1 integrin on the highgrade glioma resistance to TMZ therapy.When the α5 integrin subunit is overexpressed, the p53-mediated responses to genotoxic damage are compromised.
When the α5 integrin level is low or suppressed, p53 is stabilized and fully functional.
An inverse relationship between the α5 integrin level and p53 has been revealed through the use of the p53 activator, nutlin-3a.These results may have clinical relevance in light of the clear advantage reported here for prolonged survival of highgrade glioma patients with low α5 integrin subunit expression.
In agreement with our data, it was reported that the α5β1 integrin is overexpressed at the protein level in a significant proportion of human glioblastoma biopsies (20).Here, we show for the first time that in glioma the α5 mRNA level is negatively correlated to survival in three different cohorts of patients, which adds brain tumors to the growing list of cancers in which the α5β1 integrin should be considered as a therapeutic target.The role of p53 in TMZ resistance is far from being understood.Although several groups reported that p53 status is not predictive of response to chemotherapy with alkylating agents (18,21), more recent works suggest that the absence of a functional p53 increases TMZ sensitivity in glioma cell lines (22) and in intracranial glioblastoma xenografts (23).A trend towards an increased TMZ sensitivity in patients with p53 mutations was also suggested (24).We propose that overexpression of the α5β1 integrin in GBM represents an alternative mechanism, aside from p53 deletion/mutation, to inactivate the tumor-suppressive function of the p53 pathway.We are currently investigating the molecular mechanisms involved in the integrin-p53 crosstalk by addressing the role of α5β1 integrin in transcriptional and non-transcriptional effects of p53.Activation of the p53 pathway by nutlin-3a led to down-regulation of the α5 integrin subunit in glioma cells.Based on our data, there seems to be a cross-antagonistic interaction between the α5 integrin and p53 that was only revealed by nutlin-3a, which may explain why this drug overcomes the prosurvival activity of the integrin.Our results are similar to recent data showing that nutlin-3a downregulates the oncogene DEK or overcomes the anti-apoptotic Bcl2 overexpression thus leading to cell apoptosis (25,26).
In summary, we have shown for the first time that α5β1 integrin plays a critical role in resistance to TMZ therapy by interfering with the p53 pathway in high-grade glioma.
In addition, we have shown that activation of p53 by nutlin-3a represses the α5β1 integrin and we propose that such downregulation is an important mediator of nutlin-3a cytotoxic activity.The relevance of our results is emphasized by the finding that α5 integrin gene overexpression is associated with decreased survival in high-grade glioma patients.Our data provide the rationale for a preclinical evaluation of p53 activators and/or α5β1 integrin antagonists in a subset of high-grade glioma which express a functional p53 and high levels of α5β1 integrin.(a) U87MG and LNZ308 cell morphology after 24 hours nutlin-3a treatment.U87MG cells were transfected either with control siRNA ns or with siRNA p53 and treated with nutlin-3a (5 µM) for 24 h.Silencing of the p53 protein was verified by immunoblotting.
, 2018.© 2012 American Association for Cancer cancerres.aacrjournals.orgDownloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.Author Manuscript Published OnlineFirst on May 16, 2012; DOI: 10.1158/0008-5472.CAN-11-4199 LNZ308 cells were transfected with pcDNA ctrl or pcDNA p53 and treated with nutlin-3a (10 µM).Expression of p53 was verified by immunoblotting.Scale bars: 50 µm (b) Top, Western blot analysis of the α5 integrin protein expression in pcDNA ctrl -and pcDNA α5 -transfected U87MG cells.Cells were treated with nutlin-3a (10 µM), TMZ (200 µM) or both drugs for 24 h.Histograms show the fold increase in the protein expression normalized to GAPDH levels (mean ± sem of 3 -4 independent experiments).* p < 0.05, ** p < 0.01, ***p < 0.001 for treated cells versus non treated cells.Middle, representative fluorescence images with specific anti-α5 integrin antibodies of untreated and nutlin-treated pcDNA ctrl -and pcDNA α5 -transfected U87MG cells.Scale bars : 20 µm.Bottom, flow cytometry analysis of the α5 (IIA1 antibody) integrin subunit at the cell membrane of pcDNA ctrl -and pcDNA α5transfected U87MG cells before and after nutlin-3a (10µM) treatment for 24 hours.(c) Histograms represent the fold increase of the α5 mRNA in pcDNA ctrl -and pcDNA α5transfected U87MG cells after nutlin-3a (10 µM) treatment, * p < 0.05 for treated cells versus non treated cells.(d) Left, Western blot analysis of α5 and p53 protein expression in p53 null LNZ308 cells treated with nutlin-3a (5 and 10 µM) for 24 hours.A representative blot out of three is shown.Right, LNZ308 cells were transfected with p53wt and α5 integrin expression was detected by western blot analysis.Histogram represents the mean ± sem of 3 independent experiments.GAPDH was used as the loading control

FIGURE 4 .Figure 1 G4 (low α 5 ;Figure 1
FIGURE 4. Elevated α5-integrin gene expression is associated with high-grade glioma and predicts decreased survival rates (a) Gene expression levels of α5 and β1 integrin subunits were quantified with specific primers by qPCR in 20 non-tumor brain tissues (c), 22 grade 2 (GII), 38 Research.on January 22, 2018.© 2012 American Association for Cancer cancerres.aacrjournals.orgDownloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.Author Manuscript Published OnlineFirst on May 16, 2012; DOI: 10.1158/0008-5472.CAN-11-4199 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.Author Manuscript Published OnlineFirst onMay 16, 2012; DOI: 10.1158/0008-5472.
on January 22, 2018.© 2012 American Association for Cancer cancerres.aacrjournals.orgDownloaded from . Each sample was histologically analyzed by a Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.Author Manuscript Published OnlineFirst onMay 16, 2012; DOI: 10.1158/0008-5472.
Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.Author Manuscript Published OnlineFirst onMay 16, 2012; DOI: 10.1158/0008-5472.