Inositol-requiring enzyme 1α is a key regulator of angiogenesis and invasion in malignant glioma
Edited by Napoleone Ferrara, Genentech, Inc., South San Francisco, CA, and approved July 22, 2010 (received for review December 4, 2009)
Abstract
Inositol-requiring enzyme 1 (IRE1) is a proximal endoplasmic reticulum (ER) stress sensor and a central mediator of the unfolded protein response. In a human glioma model, inhibition of IRE1α correlated with down-regulation of prevalent proangiogenic factors such as VEGF-A, IL-1β, IL-6, and IL-8. Significant up-regulation of antiangiogenic gene transcripts was also apparent. These transcripts encode SPARC, decorin, thrombospondin-1, and other matrix proteins functionally linked to mesenchymal differentiation and glioma invasiveness. In vivo, using both the chick chorio-allantoic membrane assay and a mouse orthotopic brain model, we observed in tumors underexpressing IRE1: (i) reduction of angiogenesis and blood perfusion, (ii) a decreased growth rate, and (iii) extensive invasiveness and blood vessel cooption. This phenotypic change was consistently associated with increased overall survival in glioma-implanted recipient mice. Ectopic expression of IL-6 in IRE1-deficient tumors restored angiogenesis and neutralized vessel cooption but did not reverse the mesenchymal/infiltrative cell phenotype. The ischemia-responsive IRE1 protein is thus identified as a key regulator of tumor neovascularization and invasiveness.
Sign up for PNAS alerts.
Get alerts for new articles, or get an alert when an article is cited.
Malignant gliomas are the most frequent primary brain tumors and represent a major challenge in cancer therapy. These deadly tumors exhibit the classic features of most solid cancers, including high proliferation rate, angiogenesis, and invasiveness (1). Gliomas are characterized by diffuse infiltrative growth in the surrounding brain parenchyma, which precludes complete surgical resection and is responsible for local recurrences. A frequent feature of higher grade gliomas is their presentation of highly and abnormally vascularized lesions, a phenomenon that contributes to clinical malignancy by increasing interstitial fluid pressure, promoting vascular thrombosis followed by necrosis and occasionally inducing intracranial hemorrhages (2–4).
As a consequence of these characteristics, gliomas are not easily accessible to current therapies. Antiangiogenic treatments, in combination with chemotherapy, have proven beneficial for patients suffering from high-grade tumors (5). However, angiogenesis inhibition may also have adverse effects by triggering invasive cell behavior (6, 7). Indeed, inhibition or invalidation of key factors for angiogenesis such as the transcription factor HIF-1α (8, 9), VEGF-A and its receptor KDR (9–12), VE-cadherin (13) and MMPs (14) enhances tumor invasion and cooption of blood vessels. A transition from invasive/coopting tumors to highly angiogenic tumors, which is a hallmark of glioma progression to glioblastoma, has also been modeled in animals (15). However, the molecular mechanisms underlying these seemingly mutually exclusive behaviors have not been elucidated. Most of the above-mentioned factors are functionally linked to VEGF-A activity (3, 8), which provides a rationale for the molecular analysis of these phenotypic transitions. Expression signatures of invasive (15, 16) and angiogenic (15, 17, 18) glioma cells and tissues also constitute critical frameworks for a comprehensive approach of these pathologic events. The existence of various invasion and migration mechanisms (19) and redundant signals controlling the angiogenic switch (7) plead for therapeutic targeting of molecular functions acting at key branch points of these processes.
Inositol-requiring enzyme 1α (IRE1α, also named ERN1) is an endoplasmic reticulum (ER)-resident transmembrane protein acting as a proximal sensor of the unfolded protein response (UPR). As such, it participates in the early cellular response to the accumulation of misfolded proteins in the ER occurring under both physiological and pathological situations (20, 21). Misfolded proteins in the ER lumen activate two distinct catalytic domains of IRE1, which display serine/threonine trans-autophosphorylation and endoribonuclease (RNase) activities, respectively. IRE1-associated RNase activity is involved in the degradation of a specific subset of mRNA (22) and also initiates the cytosolic splicing of the pre-XBP1 (X-box binding protein 1) mRNA whose mature transcript encodes a transcription factor that stimulates the expression of UPR-specific genes (20, 21). Recently, single mutations in IRE1α/ERN1 gene were detected in human cancers, and IRE1 was proposed as a major contributor to tumor progression among protein kinases (23). Blockade of IRE1 transduction pathways in experimental tumor models suggests that its activity is linked to the neovascularization process (24).
We now provide evidence that inhibition of IRE1 signaling results in the reduction in the number of functional tumor blood vessels in glioma and in the increase of tumor cell invasion and vessel cooption. The general aspect of these lesions exhibited similarities with the sarcomatous component in gliosarcomas, a rare biphasic variant of glioblastomas. This phenotypic transition correlates with the coordinate modulation of a variety of pro- and anti-angiogenesis factors. Transgene expression of the cytokine IL-6 in IRE1-deficient tumors was sufficient to rescue angiogenesis and also abolished vessel cooption. We propose IRE1 as a candidate regulator of angiogenesis and invasion in malignant gliomas in relation to its property to sense ischemia.
Results
IRE1 Inhibition Induces Invasiveness and Prolongs Overall Survival of Glioma-Implanted Mice.
Inhibition of IRE1 signaling was associated to the expression of an IRE1 dominant-negative transgene in U87dn cells and was not observed in U87ctrl cells (Fig. S1). Both cell types were engrafted intracranially in mice, and tumor growth was analyzed. Surgical specimens from the U87ctrl tumors were typically homogenous and nodular with a high cellularity, round hyperchromatic nuclei and a sharp demarcation from the surrounding brain parenchyma (Fig. 1A). No significant necrosis or apoptosis was detected. This phenotype is typical for U87 cell-derived tumors (see also Fig. 2). In contrast, U87dn tumors were diffuse, highly infiltrative and revealed a sarcomatous phenotype with densely packed long bundles of spindle cells. Cells had regular shapes, loose cellular extensions characteristics of astrocytes (Fig. 1A), and further exhibited elongated hyperchromatic nuclei with prominent nucleoli. Multifocal spreading of the tumor was evident with the presence of numerous distant satellites. A dense reticulin fiber network was also observed in U87dn tumors by using the Tibor Pap silver impregnation (Fig. 1B), with every cell encased by argyrophilic filament. On the contrary, U87ctrl tumors exhibited a sparse reticulin fiber grid. The mesenchymal marker actin-α2 (sm-actin, ACTA2) was detected in mural cells of vessels in both tumor types and was also highly expressed in U87dn cells themselves, but not in U87ctrl cells (Fig. 1C and Table S1). Finally, the neural stem cell marker nestin was expressed at a higher level in U87dn cells than in U87ctrl cells (Fig. S2A). The general aspect of U87dn tumors closely resemble the sarcomatous portion of gliosarcomas (Fig. S2B; ref. 1).
Fig. 1.
Fig. 2.
The proliferation index of tumor cells was determined by using the Ki-67 labeling (Fig. 1D). Growth fraction in U87ctrl tumors was ≈37%, whereas that of U87dn tumors was ≈23%, which is in agreement with the measured volumes of the tumors and the fact that U87dn cells grew more slowly than U87ctrl cells in vitro (24). Consistently, U87wt and U87ctrl tumors exhibited much higher amounts of the human housekeeping β2-microglobulin (B2M) protein, a marker of tumor load, than U87dn tumors (Fig. S3).
We next evaluated the overall survival of mice intracerebrally engrafted with U87wt, U87ctrl, and U87dn cells. Stable transgene expression was controlled at different time points during the experimentation. Tumors developed in 100% recipient mice, independent of which cell type was implanted. However, mice engrafted with U87wt and U87ctrl cells all died within 36 d, whereas those engrafted with the different U87dn clones exhibited a significant survival extension (Fig. 1E). Markedly, mice implanted with U87 cell clones expressing angiogenic properties and higher proliferation rates (ref. 24; proliferation index, PI; PIT1P5/T2P4 > PI2A4 > PI1C5 > PI2D3) had a shorter survival. None of the U87dn tumor analyzed showed any signs of regression or apoptosis/necrosis.
Blockade of IRE1 Promotes Glioma Cell Invasion and Blood Vessel Cooption and Decreases Tumor Vascular Density and Vessel Perfusion.
Invasive U87dn tumors were further distinguished from massive U87wt and U87ctrl tumors by immunofluorescence analysis (Fig. 2). A high number of small satellite tumors were radially disposed around the U87dn primary mass (see arrowheads, Fig. 2Ab), indicating an extensive infiltration into the adjacent brain parenchyma. Migrating U87dn cells were tightly associated to preexistent blood vessels in the normal tissue (Fig. 2Ad) and were detected on the abluminal site of blood vessels (Fig. 2A e–g) as girdles of mono- or multilayered cells. In comparison, blood vessels located at the edge of U87ctrl tumors were not colonized by malignant cells (Fig. 2Ac). Thus, U87dn cells coopt and spread alongside preexisting blood vessels in healthy parenchyma. As a consequence, alteration of blood vessel morphology was apparent in regions of close contact with tumor cells (Fig. 2Ae, arrowheads) and the presence of capillary knots covered with U87dn cells nearby the invasive tip was also observed (arrows; see also ref. 25). Consistent with the cooption mechanism, a dense vascular network with numerous dilated and hyperplastic vessels was evident between tumor cell bundles. Sixty days after U87dn cell engrafting, the whole implanted hemispheres showed a widespread distribution of infiltrating tumor cells. The corpus callosum was invaded at a later stage (day ≈70), which again suggests that tumors cells preferentially migrate along blood vessels tracks.
The vascularization pattern of the two different tumor types was also strikingly distinct. As shown by CD31 labeling, high vascular density was apparent in U87ctrl tumors, and this vascularization pattern was associated with an intense angiogenesis process as demonstrated by the presence of endoglin, a marker of angiogenic tumor blood vessels (Fig. 2B). This contrasts with the nonangiogenic U87dn tumors in which large tumor areas of high cellular densities were almost completely avascular and devoided of endoglin labeling. In keeping with their invasive behavior in vivo, U87dn cells exhibited a higher ability to migrate in vitro in the presence of serum (Fig. 2C), a ≈50% higher migration rate being measured with these cells relative to U87ctrl cells. Interestingly, plasmid-directed expression of IL-6 in tumors subjected to IRE1 blockade overrode vessel cooption and rescued angiogenesis (Fig. S4), again suggesting that these two mechanisms are antagonistic in the glioma model.
We next investigated the functionality of tumor vasculature by intravital microscopy (IVM). Cells were implanted in mice either subcutaneously or intracranially, and tumor growth was observed in the two experimental settings (Fig. 2 D and E). In agreement with histological analyses, IVM revealed a significant reduction (40–95%) of the total vascular density in U87dn tumors compared with U87ctrl tumors (Fig. 2D Left). Moreover, the number of functional vessels was also much lower in IRE1dn tumors (Fig. 2D Right). Again, signs of pathological angiogenesis were observed in U87ctrl tumors (Fig. 2E), including a chaotic angio-architecture with vessels (white arrows) of heterogeneous diameters, an increase of microvascular permeability (microvessels appear in dark due to the extravasation of fluorescent dye into the tumor interstitium), and high microvascular densities. In contrast, U87dn tumor vessels could hardly be visualized because of the significant reduction of both total and functional vessels, which demonstrates a low blood perfusion in these tumors.
In Vitro Modulation of Factors Involved in Neovascularization and Invasion.
Comparison of gene and protein expression analysis of U87ctrl and U87dn cells was first examined in culture conditions on the basis of a combination of transcriptomic data, quantitative PCR (qPCR), and ELISA (Table 1, Table S1, and Fig. S5). Inhibition of IRE1 coordinately regulated factors involved in tumor angiogenesis and invasion. U87dn cells exhibited a mesenchymal profile with a higher expression level of genes encoding matrix proteins (Table S1). Besides, proangiogenic cytokines such as IL-1β, IL-6, IL-8, and VEGF-A were down-regulated in U87dn cells compared with U87ctrl cells. As a modulator of astrocytoma cell migration, SPARC mRNA expression was ≈5-fold higher in invasive U87dn cells than in noninvasive U87ctrl cells. Similarly, mRNA up-regulation of the connective tissue growth factor (CTGF), MMP-2, tissue inhibitor of metalloproteinases (TIMP)-2, decorin, thrombospondin-1 (THBS1), and perlecan (HSPG2) was also observed in U87dn tumor cells. Comparable modulation of gene expression was obtained by using a second and independent approach by siRNA-mediated knockdown of IRE1 (Fig. S6).
Table 1.
| Protein expression (fmol/106 cell per day) | |||
|---|---|---|---|
| Factor | Gene expression (dn vs. ctrl), fold change | Dn | Ctrl |
| IL-1β | 0.040 ± 0.006 | 0.6 ± 0.2 | 10 ± 1 |
| IL-6 | 0.013 ± 0.005 | 2.4 ± 0.4 | 1120 ± 160 |
| IL-8 | 0.002 ± 0.001 | 125 ± 100 | 3500 ± 500 |
| VEGF-A | 0.27 ± 0.13 | 38 ± 12 | 95 ± 17 |
| PLAU | 0.20 ± 0.04 | / | / |
| MMP-2 | 2.35 ± 0.24 | / | / |
| CTGF | 3.2 ± 0.5 | / | / |
| SPARC | 4.87 ± 1.16 | / | / |
| TIMP-2 | 10.9 ± 1.85 | / | / |
| HSPG2 | 24.91 ± 5.94 | / | / |
| decorin | 46.85 ± 0.32 | / | / |
| THBS-1 | 428 ± 164 | / | / |
Comparative gene expression analysis of the proangiogenic (U87ctrl) and proinvasive (U87dn) cells was carried out under normoxia by using qPCR. Cytokine expression rate relative to cell number was determined from cell-conditioned media by ELISA. The regulatory effect of the blockade of IRE1 on cytokines was also apparent in cells grown under hypoxia (ref. 24; Fig. S5). Results are the mean of triplicate experiments ± SD.
Angiogenic to Invasive Phenotypic Shift in the Chicken Egg Model.
The distinct features of U87ctrl and U87dn tumors were analyzed by using the chicken egg assay (12, 26). Malignant cells were deposited on the surface of the chorio-allantoic membrane (CAM), and tumor progression was observed 4 d after implantation. Striking phenotypic differences were observed between the tumor variants (Fig. 3A), which were scored according to size, degree of invasiveness, and vascularization (Fig. 3B; see criteria in SI Materials and Methods). U87wt and U87ctrl tumors were classified as large- to medium-sized (≥90% of tumors), whereas U87dn tumors were small- to medium-sized (≈80 and ≈20%, respectively). In agreement with this observation, expression of the human B2M antigen was nearly 2-fold lower in U87dn tumors than in controls (Fig. S3B). Again, U87wt and U87ctrl tumors appeared well circumscribed, highly vascularized, and noninvasive (Fig. 3 A and B), which was confirmed by immunolabeling (Fig. 3C). In contrast, U87dn tumors were devoid of blood vessels and predominantly exhibited a diffuse/invasive aspect with whitish strips arising from tumor masses and spreading along blood and lymphatic vessels inside the CAM (Fig. 3 A and C). The overall structure of the chicken membrane appeared disorganized and exhibited abnormal large-sized hyperplasic vessels to which compact tumor nodules (arrows) were closely associated. Such an invasive pattern has been described in this assay by using U87 cells knocked down for VEGF-A and IL-6 (12). Remarkably, avascular U87dn tumors developed without signs of necrosis or apoptosis.
Fig. 3.
qPCR and ELISA analyses were carried out on tumor tissues at day 4 (Fig. 3 D and E). VEGF-A and IL-6 were strongly down-regulated in U87dn tumors both at the mRNA and protein levels. Consistent with in vitro data, gene expression of CTGF, MMP-2, TIMP-2, HSPG2, decorin, and SPARC was up-regulated in avascular tumors, whereas that of MMP-1, MMP-9, and PLAU was down-modulated (Fig. 3D and Fig. S7). Similar results were obtained by using U87 cells transfected with IRE1 siRNA (Fig. S6).
Expression of Pro- and Antiangiogenic Factors in the Orthotopic Mouse Brain Model.
Gene expression was also analyzed in implanted mice brains by using laser capture microdissection (LCM) followed by qPCR (Fig. 4A). CTGF, decorin, SPARC, and THBS-1 mRNAs were consistently up-regulated in U87dn tumors in comparison with U87ctrl tumors, whereas VEGF-A, IL-6, and IL-8 mRNA expression was significantly decreased. The observed variability of these transcripts in different samples may result of the microheterogeneity of the tissue zones from which tumor samples were dissected, under the dependence of the local availability of dioxygen and nutrients (27). These results are reinforced by the fact that detected amounts of the corresponding proteins were shifted accordingly (Fig. 4 B and C). These data are consistent with those obtained both in vitro and in the CAM model.
Fig. 4.
Discussion
Angiogenesis and invasion are two manifestations of aggressive tumor behavior and depend in part on the adaptation of malignant cell to abnormal tumor microenvironments. Tumor biology is strongly influenced by ischemia, which stimulates specific molecular sensors and signal transduction machineries (28, 29). One such adaptive mechanism is the UPR, whose complex signaling network in eukaryotes regulates the flux of secretory proteins and contributes to the physiological cell response to stress (20, 21, 28). As a central component of the UPR, the IRE1/XBP1 axis is involved in cellular and tissue development and homeostasis (20, 21, 28, 30, 31). It also presents cytoprotective activities allowing adaptive responses of cancer cells to ischemic challenge and is a contributor to tumor resistance to chemotherapeutic drugs (32–34).
In the mouse brain model, blockade of IRE1 modified the mode of glioma expansion by markedly reducing angiogenesis and by promoting tumor cell invasion. Under such condition, glioma cells coopted the host vasculature and gradually infiltrated the brain along blood vessels tracks, disrupting normal tissue architecture and inducing a severe reactive gliosis (Fig. 1D).
A striking increase in survival was observed in animals implanted with invasive IRE1-dn glioma cells. The occurrence of tumor angiogenesis in malignant glioma that express functionally active IRE1 therefore shortened survival in mice, presumably by accelerating malignant growth and inducing a major mass effect. The slower progression of the dominant-negative tumors may also rely upon the intrinsic lower proliferation index of IRE1-dn cells (ref. 24; this work) in relation to their elevated capacity to migrate, because these processes are reported to be antagonistic (35).
The massive/angiogenic and diffuse/avascular phenotypes were mutually exclusive. Such a highly reproducible and stable pattern evokes a compensatory mechanism, whereby invasion manifests in response to the stringent blockade of neovascularization. A similar behavior has been described after inhibition of prominent angiogenic actors (8–13), which also present a significant association with clinical outcome in gliomas. Consistently, impairing IRE1 activity either by using a dominant-negative strategy or by siRNA knockdown led both in vitro and in vivo to the integrated modulation of several pro- and antiangiogenic factors in favor of angiogenesis inhibition. In addition to VEGF-A down-regulation, a multifold decrease of the expression of the potent angiogenic factors IL-1β, IL-6, and IL-8 was observed, therefore reducing possible redundancies and angiogenic relapses after IRE1 blockade. These data suggest that IRE1-dependent signaling is a key regulatory pathway for angiogenesis and tumor progression.
A significant body of evidence also implicates CTGF, the multivalent ECM proteins SPARC, decorin, and THBS1, and the proteolytic enzymes and inhibitors PLAU, MMP-2, and TIMP-2 in invasion and angiogenesis (2, 14, 16, 18, 36). Again, up-regulation of these genes in IRE1-dn tumors was consistent with an enhanced migratory profile. Such a gene expression pattern, in association with the reported tumor phenotypes, is indicative of a mesenchymal drift, a process by which cells lose glial and gain mesenchymal features (37). Remarkably, invasion and cooption can be overridden in U87dn tumors by ectopic expression of IL-6 and induction of angiogenesis. In these conditions, tumor cells still exhibited a mesenchymal phenotype and infiltrative properties.
Several lines of evidence indicate that IRE1 signaling is partially active in cells under basal conditions. This observation agrees with the fact that IRE1 autophosphorylation on serine 724, which is an activation signal of the protein (38), represents in standard conditions ≈60% of the total signal obtained in response to the strong UPR-inducer tunicamycin (Fig. S1). The autophosphorylation level relates to, but does not correlate with, the splicing of pre-XBP1 mRNA and up-regulation of MIST1 gene expression, which indicates a partial decoupling between IRE1 autophosphorylation event and XBP1 mRNA processing. As a single transmembrane receptor kinase, IRE1 should express a low activity in the absence of its ligands. However, inherent errors in protein biogenesis and folding occur permanently in cells. Indeed, immediate proteasome-dependent protein degradation may represent nearly 30% of the total protein production in normal tissues, and a higher degradation rate is even expected in tumor cells (39). Overexpression of secretory proteins and the increase of gene alteration and mRNA transcription infidelity, which represent common attributes of cancer cells, have predictable consequences on the quality of protein folding (28, 40). Each of these features may therefore contribute to activate at a basal level the IRE1 branch of the UPR by elevating the total fraction of proteins committed to degradation. The blockade of IRE1 activity should therefore impact on tumor cells located either in perfused or in ischemic regions. Noteworthy, IRE1 is activated under severe hypoxia (41) and expression of the cytokines IL-1β, IL-6, and IL-8 (Fig. S5), VEGF-A and CTGF as well as of the matrix proteins THBS1 and MMP-2 and -9 is appropriately modulated under low oxygen (29, 36, 42). Activity of IRE1 is therefore likely to be intensified in an ischemic tumor microenvironment.
Angiogenesis inhibition has emerged as a promising therapy for both newly diagnosed and recurrent tumors. However, antiangiogenesis treatments have limited efficacy in some tumors and may also lead to adverse effects by inducing a highly invasive and metastatic behavior. In line with these results, we observed that a widespread dissemination of glioma cells in brain tissues occurred in response to the suppression of the upstream UPR sensor IRE1 and concomitant inhibition of tumor neovascularization. Such a protumorigenic cell adaptation was also considered in preclinical and clinical studies linking inhibition of glioma angiogenesis to the increase of cancer cells invasion (5–7). Particular attention should also be paid to the potential protooncogenic effects of IRE1 mutations (23) and to the inhibition of IRE1-mediated apoptosis (28) in the development of anti-IRE1 strategies useful for the clinic. Selective pharmacological targeting of IRE1 kinase- or RNase-dependent signaling pathways, instead of the whole IRE1 protein itself, may therefore represent a more appropriate approach for therapy. Such a strategy may present an additional interest by weakening tumor cell responses to conventional chemotherapeutic treatments (32–34). A comprehensive analysis of IRE1-dependent signaling is required for the understanding of the relationship between angiogenesis and invasiveness to better circumvent tumor cell escape to angiogenesis inhibitors.
Materials and Methods
Reagents.
Antibodies against the following antigens were used as follows: GFAP, Ki-67, and sm-actin were from DAKO; mouse CD31 and mouse endoglin (B&D Pharmingen); vimentin Ab-2 (Interchim); SPARC (US Biological); nestin (Millipore); phospho-(Ser724)-IRE1 (Abcam); and IRE1 and β-actin (Santa Cruz Biotechnology). Secondary antibodies labeled with FITC, AlexaFluor488, or AlexaFluor546 were from Invitrogen. SNA1-lectin coupled to FITC was from Vector Laboratories. Primers (Table S2) were from Sigma Aldrich.
Cell Culture Experiments.
Wild-type U87-MG (U87wt) glioma cells (ATCC; HTB-14), empty plasmid U87 cells (U87ctrl; clones T1P5 and T2P4), U87 IRE1.NCK DN cells (U87dn; clones 1C5, 2A4, and 2D3), and iC6 clones were grown as described (24). Small interfering RNA knockdown experiments and cell migration assay are described in SI Materials and Methods.
CAM Assay.
Three million U87 cells were deposited in 15 μL of DMEM onto the chicken CAM at day 10 of development (26). At day 4 after implantation, tumors were excised and pooled (n = 5 for each condition) before RNA and protein extraction. Tumor replicates (n ≥ 3) were also cut into 10-μm-thick cryosections before immuno-labeling. Classification of tumor phenotypes was established as reported in SI Materials and Methods.
Intracranial Implantations.
Implantations of U87 cells were performed in 8–9 wk of age RAG2/γc mice (24). Mice were killed according to defined ethical criteria and were perfused with 4% neutral buffered formalin before brains removal. Brains were kept overnight in sucrose 15% and embedded in tissue-tek O.C.T. for cryosection.
Histopathological and Immunohistochemical Analyses.
IHC was performed either on paraffin-embedded sections or on cryosections as reported in SI Materials and Methods.
LCM Analyses.
Frozen sections (30 μm) of mouse brains were obtained and were mounted on PEN-membrane 1-mm glass slides. Tissue sections were fixed in ethanol, stained with H&E, dehydrated, and air-dried. LCM was performed by using a PALM MicroBeam microdissection system version 4.0–1206 equipped with a P.A.L.M. RoboSoftware (P.A.L.M. Microlaser Technologies). Four tumors were analyzed for each condition, and five caps were collected for each tumor type. RNA samples with an RNA integrity number greater than 8 were selected after Bioanalyzer validation.
IVM.
Intravital-microscopic assessment of tumor microcirculation was performed in nude mice as described (43, 44). Details of the procedure are reported in SI Materials and Methods.
RNA and Protein Analyses.
RT-PCR, PCR, ELISA, and immunoblotting procedures are described in SI Materials and Methods.
Gene Expression Profile.
cRNAs from subconfluent U87ctrl and U87dn cells in culture were generated from 10 μg of total RNA and hybridized to Human U133A 2.0 chips (Affymetrix) at the Affymetrix transcriptome platform (12). Each cell type was used in three separate experiments. Analysis of CEL files was performed by using the Limma package available through Bioconductor (45) with R statistical software and the OneChannelGUI graphical interface (Table S1). Probe set intensities were obtained by means of GCRMA and were selected by using a corrected P value treshhold of 0.05 and fold change treshhold of |log2(fc)| ≥ 1.
Data Availability
Data deposition: The data reported in this paper have been deposited in the Gene Expression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE22385).
Acknowledgments
We are grateful to D. Jaeger (Department of Neuropathology, University of Heidelberg) for assistance in IHC stainings; M. Hagedorn, S. Javerzat, and M. Franco for their contribution to the blind test in the CAM assay; G. Parmaksiz and S. Bayerl for assistance in performing IVM experiments; and N. Platonova and F. Delhaes for help with IL-6 expression analysis. This work was supported by Ministère de l'Éducation Nationale de l'Enseignement Supérieur et de la Recherche, Institut National de la Santé et de la Recherche Médicale, and by grants from the Ligue Nationale Contre le Cancer, Comité de la Gironde (M. Moenner); Association pour la Recherche sur le Cancer Grants 3694 and 1097 (to A.B. and M. Moenner, respectively); Institut National du Cancer (Gliostress) (A.B.); and an IFR66 grant (to E.C. and M. Moenner). The LCM platform at Institut National de la Santé et de la Recherche Médicale U862 was financed by the Fondation pour la Recherche Médicale. G.A. is a postdoctoral fellow of Bonus Qualité Recherche, University Bordeaux 1, and was also supported by the European Consortium for Tumor Angiogenesis Research (Angiotargeting).
Supporting Information
Supporting Information (PDF)
Supporting Information
- Download
- 1.05 MB
References
1
P Kleihues, et al., Glioblastoma. WHO Classification of Tumours of the Central Nervous System, eds DN Louis, H Ohgaki, OD Wiestler, WK Cavenee (Intl Agency Res Cancer, Lyon, France), pp. 33–49 (2007).
2
P Carmeliet, RK Jain, Angiogenesis in cancer and other diseases. Nature 407, 249–257 (2000).
3
RK Jain, et al., Angiogenesis in brain tumours. Nat Rev Neurosci 8, 610–622 (2007).
4
G Lee, AC Zacest, PL Reilly, Recurrent haemorrhage from a malignant brain stem glioma: A review of the mechanisms of tumoural haemorrhage. J Clin Neurosci 10, 632–634 (2003).
5
JF de Groot, et al., Tumor invasion after treatment of glioblastoma with bevacizumab: Radiographic and pathologic correlation in humans and mice. Neuro-oncol 12, 233–242 (2010).
6
K Lamszus, P Kunkel, M Westphal, Invasion as limitation to anti-angiogenic glioma therapy. Acta Neurochir Suppl (Wien) 88, 169–177 (2003).
7
S Loges, M Mazzone, P Hohensinner, P Carmeliet, Silencing or fueling metastasis with VEGF inhibitors: Antiangiogenesis revisited. Cancer Cell 15, 167–170 (2009).
8
B Blouw, et al., The hypoxic response of tumors is dependent on their microenvironment. Cancer Cell 4, 133–146 (2003).
9
R Du, et al., HIF1alpha induces the recruitment of bone marrow-derived vascular modulatory cells to regulate tumor angiogenesis and invasion. Cancer Cell 13, 206–220 (2008).
10
M Pàez-Ribes, et al., Antiangiogenic therapy elicits malignant progression of tumors to increased local invasion and distant metastasis. Cancer Cell 15, 220–231 (2009).
11
JL Rubenstein, et al., Anti-VEGF antibody treatment of glioblastoma prolongs survival but results in increased vascular cooption. Neoplasia 2, 306–314 (2000).
12
A Saidi, et al., Combined targeting of interleukin-6 and vascular endothelial growth factor potently inhibits glioma growth and invasiveness. Int J Cancer 125, 1054–1064 (2009).
13
K Lamszus, et al., Inhibition of glioblastoma angiogenesis and invasion by combined treatments directed against vascular endothelial growth factor receptor-2, epidermal growth factor receptor, and vascular endothelial-cadherin. Clin Cancer Res 11, 4934–4940 (2005).
14
R Du, et al., Matrix metalloproteinase-2 regulates vascular patterning and growth affecting tumor cell survival and invasion in GBM. Neuro-oncol 10, 254–264 (2008).
15
PO Sakariassen, et al., Angiogenesis-independent tumor growth mediated by stem-like cancer cells. Proc Natl Acad Sci USA 103, 16466–16471 (2006).
16
T Demuth, et al., Glioma cells on the run—the migratory transcriptome of 10 human glioma cell lines. BMC Genomics 9, 54 (2008).
17
N Almog, et al., Transcriptional switch of dormant tumors to fast-growing angiogenic phenotype. Cancer Res 69, 836–844 (2009).
18
S Godard, et al., Classification of human astrocytic gliomas on the basis of gene expression: A correlated group of genes with angiogenic activity emerges as a strong predictor of subtypes. Cancer Res 63, 6613–6625 (2003).
19
P Friedl, K Wolf, Tumour-cell invasion and migration: Diversity and escape mechanisms. Nat Rev Cancer 3, 362–374 (2003).
20
D Ron, P Walter, Signal integration in the endoplasmic reticulum unfolded protein response. Nat Rev Mol Cell Biol 8, 519–529 (2007).
21
M Schröder, RJ Kaufman, ER stress and the unfolded protein response. Mutat Res 569, 29–63 (2005).
22
J Hollien, et al., Regulated Ire1-dependent decay of messenger RNAs in mammalian cells. J Cell Biol 186, 323–331 (2009).
23
C Greenman, et al., Patterns of somatic mutation in human cancer genomes. Nature 446, 153–158 (2007).
24
B Drogat, et al., IRE1 signaling is essential for ischemia-induced vascular endothelial growth factor-A expression and contributes to angiogenesis and tumor growth in vivo. Cancer Res 67, 6700–6707 (2007).
25
F Winkler, et al., Imaging glioma cell invasion in vivo reveals mechanisms of dissemination and peritumoral angiogenesis. Glia 57, 1306–1315 (2009).
26
M Hagedorn, et al., Accessing key steps of human tumor progression in vivo by using an avian embryo model. Proc Natl Acad Sci USA 102, 1643–1648 (2005).
27
D Shweiki, M Neeman, A Itin, E Keshet, Induction of vascular endothelial growth factor expression by hypoxia and by glucose deficiency in multicell spheroids: Implications for tumor angiogenesis. Proc Natl Acad Sci USA 92, 768–772 (1995).
28
Y Ma, LM Hendershot, The role of the unfolded protein response in tumour development: Friend or foe? Nat Rev Cancer 4, 966–977 (2004).
29
CW Pugh, PJ Ratcliffe, Regulation of angiogenesis by hypoxia: Role of the HIF system. Nat Med 9, 677–684 (2003).
30
D Acosta-Alvear, et al., XBP1 controls diverse cell type- and condition-specific transcriptional regulatory networks. Mol Cell 27, 53–66 (2007).
31
NN Iwakoshi, AH Lee, LH Glimcher, The X-box binding protein-1 transcription factor is required for plasma cell differentiation and the unfolded protein response. Immunol Rev 194, 29–38 (2003).
32
AH Lee, NN Iwakoshi, KC Anderson, LH Glimcher, Proteasome inhibitors disrupt the unfolded protein response in myeloma cells. Proc Natl Acad Sci USA 100, 9946–9951 (2003).
33
ST Nawrocki, et al., Bortezomib sensitizes pancreatic cancer cells to endoplasmic reticulum stress-mediated apoptosis. Cancer Res 65, 11658–11666 (2005).
34
M Rahmani, et al., The kinase inhibitor sorafenib induces cell death through a process involving induction of endoplasmic reticulum stress. Mol Cell Biol 27, 5499–5513 (2007).
35
A Giese, et al., Dichotomy of astrocytoma migration and proliferation. Int J Cancer 67, 275–282 (1996).
36
AC Bellail, SB Hunter, DJ Brat, C Tan, EG Van Meir, Microregional extracellular matrix heterogeneity in brain modulates glioma cell invasion. Int J Biochem Cell Biol 36, 1046–1069 (2004).
37
CL Tso, et al., Primary glioblastomas express mesenchymal stem-like properties. Mol Cancer Res 4, 607–619 (2006).
38
KL Lipson, et al., Regulation of insulin biosynthesis in pancreatic beta cells by an endoplasmic reticulum-resident protein kinase IRE1. Cell Metab 4, 245–254 (2006).
39
U Schubert, et al., Rapid degradation of a large fraction of newly synthesized proteins by proteasomes. Nature 404, 770–774 (2000).
40
M Brulliard, et al., Nonrandom variations in human cancer ESTs indicate that mRNA heterogeneity increases during carcinogenesis. Proc Natl Acad Sci USA 104, 7522–7527 (2007).
41
C Koumenis, BG Wouters, “Translating” tumor hypoxia: Unfolded protein response (UPR)-dependent and UPR-independent pathways. Mol Cancer Res 4, 423–436 (2006).
42
KL Bennewith, et al., The role of tumor cell-derived connective tissue growth factor (CTGF/CCN2) in pancreatic tumor growth. Cancer Res 69, 775–784 (2009).
43
M Czabanka, M Vinci, F Heppner, A Ullrich, P Vajkoczy, Effects of sunitinib on tumor hemodynamics and delivery of chemotherapy. Int J Cancer 124, 1293–1300 (2009).
44
P Vajkoczy, L Schilling, A Ullrich, P Schmiedek, MD Menger, Characterization of angiogenesis and microcirculation of high-grade glioma: An intravital multifluorescence microscopic approach in the athymic nude mouse. J Cereb Blood Flow Metab 18, 510–520 (1998).
45
RC Gentleman, et al., Bioconductor: Open software development for computational biology and bioinformatics. Genome Biol 5, R80 (2004).
Information & Authors
Information
Published in
Classifications
Copyright
Freely available online through the PNAS open access option.
Data Availability
Data deposition: The data reported in this paper have been deposited in the Gene Expression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE22385).
Submission history
Published online: August 11, 2010
Published in issue: August 31, 2010
Keywords
Acknowledgments
We are grateful to D. Jaeger (Department of Neuropathology, University of Heidelberg) for assistance in IHC stainings; M. Hagedorn, S. Javerzat, and M. Franco for their contribution to the blind test in the CAM assay; G. Parmaksiz and S. Bayerl for assistance in performing IVM experiments; and N. Platonova and F. Delhaes for help with IL-6 expression analysis. This work was supported by Ministère de l'Éducation Nationale de l'Enseignement Supérieur et de la Recherche, Institut National de la Santé et de la Recherche Médicale, and by grants from the Ligue Nationale Contre le Cancer, Comité de la Gironde (M. Moenner); Association pour la Recherche sur le Cancer Grants 3694 and 1097 (to A.B. and M. Moenner, respectively); Institut National du Cancer (Gliostress) (A.B.); and an IFR66 grant (to E.C. and M. Moenner). The LCM platform at Institut National de la Santé et de la Recherche Médicale U862 was financed by the Fondation pour la Recherche Médicale. G.A. is a postdoctoral fellow of Bonus Qualité Recherche, University Bordeaux 1, and was also supported by the European Consortium for Tumor Angiogenesis Research (Angiotargeting).
Notes
This article is a PNAS Direct Submission.
Authors
Competing Interests
The authors declare no conflict of interest.
Metrics & Citations
Metrics
Citation statements
Altmetrics
Citations
Cite this article
107 (35) 15553-15558,
Export the article citation data by selecting a format from the list below and clicking Export.
Cited by
Loading...
View Options
View options
PDF format
Download this article as a PDF file
DOWNLOAD PDFLogin options
Check if you have access through your login credentials or your institution to get full access on this article.
Personal login Institutional LoginRecommend to a librarian
Recommend PNAS to a LibrarianPurchase options
Purchase this article to access the full text.