Periostin (POSTN) interacts with multiple integrins to coordinate a variety of cellular processes, including epithelial-to-mesenchymal transition (EMT) and cell migration. In our previous study, anti-VEGF-A therapy was associated with resistance and EMT. This study sought to determine the role of POSTN in the resistance of glioma stem cells (GSC) to antiangiogenic therapy. In mouse xenograft models of human glioma, POSTN expression was associated with acquired resistance to anti-VEGF-A therapy and had a synergistic effect with bevacizumab in prolonging survival and decreasing tumor volume. Resistance to anti-VEGF-A therapy regulated by POSTN was associated with increased expression of TGFβ1 and hypoxia-inducible factor-1α (HIF1α) in GSCs. At the molecular level, POSTN regulated invasion and expression of EMT (caveolin-1) and angiogenesis-related genes (HIF1α and VEGF-A) through activation of STAT3. Moreover, recombinant POSTN increased GSC invasion. Collectively, our findings suggest that POSTN plays an important role in glioma invasion and resistance to antiangiogenic therapy. Mol Cancer Ther; 15(9); 2187–97. ©2016 AACR.

This article is featured in Highlights of This Issue, p. 2009

Tumor angiogenesis is regulated by angiogenic factors, including VEGF, platelet-derived growth factor, and hypoxia-inducible factor-1α (HIF1α; ref. 1). The transcription factor HIF1α plays a critical role in regulation of VEGF stabilization during hypoxia (2). Glioma stem cells (GSC) increase tumor angiogenesis via elevated secretion of VEGF-A (3). The production of VEGF in gliomas is significant, with researchers finding VEGF levels 200- to 300-fold higher in cystic fluid than in serum in glioblastoma patients (4). Glioblastoma is the most malignant brain tumor and is highly resistant to intensive combination therapies (5). Furthermore, glioblastomas are highly vascularized tumors, and high microvessel densities in gliomas are highly correlated with poor prognosis (6). Proliferation of endothelial cells is observed frequently and exclusively in mesenchymal-type glioblastomas, correlating with poor prognosis (7). Bevacizumab is a humanized recombinant mAb against VEGF-A that is composed of human IgG1 constant and murine VEGF-binding regions (8). Studies of antiangiogenic therapies in xenograft mouse models of human glioblastoma encouraged clinical studies of anti-VEGF-A antibodies (9). Resistance of glioma to anti-VEGF-A therapy is associated with the epithelial-to-mesenchymal transition (EMT) (10). Besides their roles in tumor cell invasion, these alternative neovascularization mechanisms may contribute to resistance of glioblastoma to antiangiogenic therapies (5).

Periostin (POSTN; osteoblast-specific factor 2) is a 90-kDa extracellular matrix (ECM) protein containing an amino-terminal EMI domain, tandem repeats of four fasciclin domains, and a carboxy-terminal (C-terminal) domain, including a heparin-binding site (11). In humans, fasciclin I domains are found in βigh3 (12), stabilin (13), and POSTN (14). Stromal POSTN plays a key role in regulation of cancer stem cell maintenance and expansion during metastatic colonization (15). It was reported that POSTN functions as a progression-associated and prognostic biomarker in glioma cases via induction of an invasive and proliferative phenotype (16). In another study, POSTN mRNA expression was markedly higher in grade IV gliomas than in grade II and III tumors (17). POSTN interacts with several integrin receptors, such as αvβ1 and αvβ3, to regulate cellular responses such as cell proliferation, EMT, and cell migration (18). During EMT, POSTN can increase vimentin, fibronectin, and matrix metalloproteinase (MMP)-9 expression (19). Moreover, POSTN increases angiogenesis via VEGFR2 expression in endothelial cells (20), and upregulation of VEGF-C expression may promote lymphangiogenesis (21). POSTN is induced by TGFβ (22) and BMP2 (23), which has been detected in glioma cell cultures and in cerebrospinal fluid and brain tumor biopsy samples from glioma patients (24). Also, glioblastomas release active TGFβ1 and TGFβ2 (25). Although POSTN is known to be important to glioma progression and to VEGF-A expression in other cancer cells, its role in these processes and its relationship to TGFβ are unclear. Our purpose here was to determine the relationship of POSTN to invasion and VEGF-A expression in GSCs and its role in response to VEGF-A therapy in in vivo glioma models.

Cell lines, reagents, and treatment

GSC lines were obtained from Drs. Howard Colman, Erik Sulman, and Frederick Lang (The University of Texas MD Anderson Cancer Center, Houston, TX). GSC lines were isolated from fresh surgical specimens of human glioblastoma and cultured as glioblastoma neurospheres in neurosphere medium: DMEM-F12 medium (1:1) with added B27 (Invitrogen), basic fibroblast growth factor (bFGF; Sigma), and EGF (Sigma; ref. 26). Acquisition of these cell lines was approved by the Institutional Review Board of MD Anderson Cancer Center and were obtained from 2005 to 2012. 293T embryonic kidney cells were obtained from the ATCC and maintained in DMEM (Sigma) supplemented with 10% FBS. All cell lines were tested and authenticated by DNA typing at the MD Anderson Cancer Center Cell Line Characterization Core and were subsequently verified for our study (December 2014). GSC lines were reported previously (10). All cell lines were cultured early-passage at 37°C in a humidified atmosphere of 5% CO2 and 95% air. All cell lines were free of mycoplasma contamination. Recombinant POSTN (cat. #3548-F2-050) and TGFβ1 (cat. #240-B-010) were purchased from R&D Systems.

Stable knockdown of POSTN in GSCs

293T cells were transfected with short hairpin RNA (shRNA) packaged with pMD2G and pCMVR8.74 DNA using PolyJet reagents (cat. #SL100688; Signagen) according to the manufacturer's instructions. Empty vector (pGIPZ control, cat. #RHS4349), POSTN shRNA1 (cat. #V3LHS_363450), and POSTN shRNA2 (cat. #V3LHS_363449) were purchased from GE Healthcare. Negative (scramble) shRNA was purchased from Sigma (cat. #SHC002). GSCs were infected with the 293T viral soup and selected with puromycin (1 μg/mL) for 7 days. Each well of a 96-well plate was seeded with a single cell, and the high-efficiency cells were selected for further analysis.

Animal xenografts

For the in vivo experiments, we used 4- to 6-week-old female nude mice strictly inbred at MD Anderson Cancer Center and maintained in the MD Anderson Research Animal Support Facility in accordance with Institute of Laboratory Animal Resources standards. The cells (5 × 105) were implanted intracranially into the nude mice as described previously (10). Four days after implantation, bevacizumab [Avastin (10 mg/kg); Genentech] or vehicle was administered via intraperitoneal injection twice a week. The mice were killed at the indicated intervals, and their brains were removed and processed for analysis. Tumor volumes were calculated by the formula [(width) × (width) × (length)]/2. The animal experiments were approved by the MD Anderson Institutional Animal Care and Use Committee.

Gene expression analysis

Total RNA was extracted from GSCs using an RNeasy Mini Kit coupled with DNase (QIAGEN) and reverse-transcribed using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Gene expression was analyzed by using RT2 Profiler PCR arrays (SA Biosciences) as per the manufacturer's instructions. EMT (cat. PAHS-090Z) and angiogenesis (cat. PAHS-024Z) PCR arrays were used. Quantitative PCR was performed according to detection protocols recommended by SA Biosciences. Ct values for a reaction set were analyzed online at the manufacturer's website.

Microarray

Fresh frozen tumor tissue (after implantation 7 weeks) was processed and total RNA extracted according to standard methods (Qiagen) for transcriptome analysis using Affymetrix GeneChipHuman GenomeHG-U133 Plus 2.0 Arrays (cat. #900466; Affymetrix). Quality of samples was confirmed and microarray processing were carried out by the MD Anderson Cancer Center Sequencing and Microarray Core Facility. The microarray data have been submitted to the Gene Expression Omnibus (GEO) public database at NCBI, accession number GSE73071.

Western blot analysis

GSCs were subjected to lysis in radio-immunoprecipitation assay lysis buffer (Cell Signaling Technology) containing proteinase (Sigma-Aldrich) and a phosphatase cocktail (Thermo Fisher Scientific). The protein concentration in each supernatant was determined by a bicinchoninic acid protein assay (Bio-Rad). Samples were subjected to SDS-polyacrylamide gel separation, and the separated proteins were electrophoretically transferred to polyvinylidene fluoride membranes. Blots were incubated with the primary antibody overnight at 4°C and incubated with horseradish peroxidase–linked secondary anti-rabbit or anti-mouse antibody (Bio-Rad). Antibodies against (STAT3; cat. #9139), phosphorylated STAT3 (Tyr705; cat. #9135), caveolin-1 (cat. #3238), N-cadherin (cat. #4061), SMAD2 (cat. #5339), SMAD3 (cat. #9523), phosphorylated SMAD2 (cat. #3108), and phosphorylated SMAD3 (cat. #9520) were purchased from Cell Signaling Technology. Other antibodies used for Western blotting were POSTN (cat. #AP11962b; Abgent); HIF1α (cat. #610958; BD Biosciences); integrin β1 (cat. #18887), integrin β3 (cat. #6627), and GAPDH (cat. #32233; all, Santa Cruz Biotechnology); and α-tubulin (cat. #T9026; Sigma-Aldrich).

IHC and immunofluorescence

Single cells were plated on precoated poly-l-lysine coverslips and fixed in 1% paraformaldehyde for 10 minutes, rinsed with PBS solution at least three times, blocked in 10% goat serum with 0.2% Triton X-100 for 1 hour, and washed at least three times with PBS and 0.2% Triton X-100. Brain tissues were fixed in 4% paraformaldehyde for 24 hours, embedded in paraffin, sectioned serially (4 μm), and stained with hematoxylin and eosin (H&E; Sigma-Aldrich). For immunohistochemical staining, slides were deparaffinized and subjected to graded rehydration. After blocking in 5% serum and antigen retrieval citrate buffer (pH 6.0), the slides were incubated with the primary antibodies overnight at 4°C. After the slides were washed in PBS with Tween 20, the primary antibody reactions were detected by using a Vectastain ABC Kit (Vector Laboratories) with the respective secondary antibody. For immunofluorescence studies, tissue sections, after blocking, were incubated with an antibody overnight at 4°C, and then with secondary antibody (Invitrogen) for 1 hour at room temperature. Antibodies used included F4/80 (cat. #14-4801; eBioscience), POSTN [cat. #49480 (Santa Cruz Biotechnology); cat. #Ap11962b (Abgent)], nestin [cat. #Ab6142 (Abcam); cat. #2027 (Santa Cruz Biotechnology)], TGFβ1 (cat. #ab64715; Abcam), and HIF1α (cat. #610958; BD Biosciences).

Invasion assay

A Matrigel basement membrane matrix (BD Biosciences) was used to conduct in vitro cell invasion assays. Transwell inserts in 24-well plates were coated with diluted Matrigel, and cells were added to the transwells in triplicate. The cells (2 × 104/200 μL), with specific antibody [mouse IgG, integrin β1 AB cat. #MAB17781 (R&D Systems) or integrin β3 AB cat. #MAB3050 (R&D Systems)], were added to the top part of each well. Serum-free medium containing recombinant TGFβ1 or POSTN was added to the bottom of each well. Plates were incubated at 37°C, and the transwell filters were fixed and stained with 0.1% crystal violet in 20% methanol. Invasive cells were visualized using bright-field microscopy. Transwell membranes were then incubated with 2% deoxycholic acid for 20 minutes, and the absorbance of invaded cells at 595 nm was recorded.

ELISAs

Cell culture supernatants were collected and concentrations of VEGF-A and POSTN determined by human VEGF-A–specific (R&D Systems) and POSTN-specific (RayBiotech) ELISA kits according to the manufacturers' instructions. Mean concentrations were recorded in picograms per milliliter.

Detection of multiple signaling pathways using reverse-phase protein array

We used reverse-phase protein array analysis (RPPA) to detect activated signaling pathways in GSCs and gliomas extracted from mice. Cells and tissues were prepared for RPPA analysis by recommended protocols, and samples were probed with 218 and 166 validated primary antibodies, respectively, at the MD Anderson Functional Proteomics Reverse Phase Protein Array Core facility.

Statistical analysis

Unless otherwise noted, all reported results are from three independent experiments performed in triplicate. All statistical analyses were conducted by the InStat software program for Microsoft Windows (GraphPad Software). Data are reported as the mean ± SD. Survival was analyzed by the Kaplan–Meier method. Tumor volumes among groups were compared using the log-rank test. All other data were compared by the unpaired two-tailed Student t test.

Glioma resistance to anti-VEGF-A therapy is associated with POSTN expression

A previous study found that anti-VEGF-A therapy induced increases in invasive characteristics in both GSC11 and U87 murine tumors while increasing median overall host survival (27). In this study, we investigated the relationship between POSTN expression and bevacizumab resistance in murine bearing GSCs. In our previous model of anti-VEGF-A resistance (27), murine GSC11 and U87 tumors treated with bevacizumab expressed higher POSTN levels than control tumors (Fig. 1A). We therefore examined the effect of POSTN in vivo on response to anti-VEGF-A therapy in mice. The median survival duration in animals implanted with control GSC272 cells was 70 days; median survival in animals implanted with GSC272 cells and treated with bevacizumab was 93 days; and median survival in animals implanted with POSTN-knockdown #1 and #2 GSC272 were 85 and 97 days, respectively (Fig. 1B). As expected, and consistently with a previous study (28), bevacizumab-based treatment resulted in much longer survival durations than in control mice. Moreover, knockdown of POSTN expression increased survival durations than in treatment with bevacizumab. Interestingly, the combination of POSTN knockdown and treatment with bevacizumab increased the median survival duration from 70 days to 100.5 (shRNA #1) and 115 (shRNA#2) days, respectively. After 7 weeks of treatment, the mean tumor volume in mice treated with bevacizumab (0.05 mm3) and that in mice given POSTN shRNA–infected GSC272 cells (shRNA#1: 1.35; shRNA#2: 1.5 mm3) were smaller than in control mice (9.63 mm3). However, the group given POSTN shRNA GSC272 cells and treated with bevacizumab had no tumors (Fig. 1C). After 17 weeks of treatment, only the mice given this combination were still alive (Fig. 1B). At 16 weeks, the mean tumor volume in mice given the combination (shRNA#1: 30.79; shRNA#2: 25.65 mm3) was lower than those in the mice treated with bevacizumab (100 mm3) or POSTN shRNA GSC272 cells (shRNA#1: 70.35; shRNA#2: 61.06 mm3). In mice implanted with GSC11 cells, those given the combination of POSTN-knockdown GSC11 cells with bevacizumab treatment had lower tumor volumes than the mice implanted with parental GSC11 cells given bevacizumab (P < 0.01; Supplementary Fig. S1A). In particular, the mean number of nestin-positive cells in mice given POSTN shRNA-infected GSC11 cells alone or POSTN shRNA following bevacizumab were lower than those in the control and bevacizumab-treated mice (Supplementary Fig. S1B). Immunohistochemical staining of GSC272 tumor samples found that POSTN expression was much higher in bevacizumab-treated mice than in control mice. POSTN expression was not detectable in mice given POSTN shRNA-infected GSCs regardless of bevacizumab treatment (Fig. 1D).

POSTN expression in GSCs regulates TGFβ1 expression in macrophages

To determine the location of the expression of POSTN in GSCs relative to that in macrophages, we double-stained the stem cell marker nestin or the macrophage marker F4/80. We observed that POSTN was double-stained with nestin but not with F4/80 (Fig. 2A and Supplementary Fig. S2A). Others have found that TGFβ1 promoted the glioma mesenchymal phenotype and that its expression was markedly increased in a bevacizumab-resistant GSC cell line (10). Treatment with bevacizumab in this study increased TGFβ1 expression in murine GSC272 and GSC11 xenografts (Fig. 2B; Supplementary Fig. S2B). The groups given both POSTN shRNA-infected GSCs and the combination of POSTN shRNA and bevacizumab did not have TGFβ1 expression. TGFβ1–expressing murine GSC272 tumors treated with bevacizumab exhibited double-staining for F4/80 (Fig. 2C). In addition, expression of HIF1α, a key regulator of resistance to anti-VEGF-A therapy, was lower in the POSTN shRNA groups than in the control (Fig. 2D). These findings demonstrate that POSTN was expressed in GSCs and regulated the expression of TGFβ1 and HIF1α during anti-VEGF-A therapy.

Expression of POSTN regulates invasion and VEGF-A expression in GSCs

To investigate the regulation of POSTN expression in GSCs, we performed Western blot analyzed the POSTN expression of several GSC lines. Expression of POSTN was highest in GSC6-27, GSC11, and GSC272 cells (Fig. 3A). Stable knockdown of POSTN expression by the shRNA1, 2, and negative shRNA construct in GSCs abrogated POSTN expression (Supplementary Fig. S3A). We compared the EMT and angiogenesis profiles of vector- and POSTN shRNA#2-infected GSC272 cells using a PCR-based arrays (Supplementary Fig. S3B). The results of this analysis are summarized in Supplementary Table S1. To expand on these findings, we performed VEGF-A ELISA and invasion assays. As shown in Fig. 3B, VEGF-A expression decreased in POSTN shRNA–infected GSC272 cells. Also, POSTN shRNA infection decreased invasion of both GSC11 and GSC272 cells (Fig. 3B and Supplementary Fig. S3C). HIF1α is key regulation factor for VEGF-A, so we performed an analysis of expression of HIF1α and phosphorylation of STAT3 in GSCs. Consistent with VEGF-A expression, expression of HIF1α and phosphorylation of STAT3 decreased following knockdown of POSTN expression (Fig. 3C). EMT-related genes, N-cadherin and caveolin-1 expression were decreased by POSTN knockdown (Supplementary Fig. S3D). To elucidate the role of STAT3 in POSTN regulation, we examined the effect of a STAT3 inhibitor (AZD1480) on HIF1α and caveolin-1 expression. Treatment with the STAT3 inhibitor decreased levels of HIF1 α, VEGF-A, and caveolin-1 compared with control treatment (Supplementary Fig. S3E). Also, expression of POSTN in GSC272 was decreased by the STAT3 inhibitor.

TGFβ1 increase POSTN secretion and downstream signaling

Growth factors such as TGFβ, EGF, and hepatocyte growth factor activate EMT (29–32), and primary lung fibroblasts upregulate POSTN expression in response to TGFβ3 and TGFβ2 (15). To identify the mechanism underlying regulation of POSTN expression by TGFβ1, we treated GSC272 cells with TGFβ1. TGFβ1 increased the secretion of POSTN and phosphorylation of SMAD3 (Fig. 4A) and SMAD2 (Supplementary Fig. S4A). After treatment with TGFβ1, secretion of POSTN increased greatly at 1 and 24 hours. Also, secretion of POSTN was blocked by an inhibitor of TGFβ1 receptor with or without TGFβ1 stimulation (Supplementary Fig. S4A). We found, furthermore, that TGFβ1 increased the expression of HIF1α, VEGF-A, and STAT3 in GSC272 and the invasion via POSTN (Fig. 4B and C and Supplementary Fig. S4B).

Recombinant TGFβ1 and POSTN stimulate GSC invasion via integrin β1

Integrins influence cells' interaction with POSTN. Therefore, we characterized several integrin receptors in GSC lines. GSC11 cells expressed integrin β1 and integrin β3, whereas GSC272 cells expressed only integrin β1 (Fig. 5A). In our evaluation of the role of the integrin β1 and integrin β3 receptors in POSTN expression, an anti-integrin β1 antibody blocked invasion induced by recombinant TGFβ1 and POSTN, whereas integrin β3 had no effect on invasion (Fig. 5B and Supplementary Fig. S5). Furthermore, recombinant POSTN stimulated phosphorylation of STAT3 and expression of caveolin-1. POSTN knockdown decreased expression of STAT3 and caveolin-1 protein induced by recombinant POSTN (Fig. 5C).

Identification of POSTN-regulated genes in vitro and in vivo

To examine other potential markers of response of GSCs to POSTN, we used microarray analysis and RPPA. Differential mRNA expression patterns were seen in GSC272-vector compared with GSC272-POSTN shRNA tumors (Supplementary Fig. S6). Likewise, RPPA data demonstrated that expression levels of pAKT (pT308, pS473), pS6 (S235, S236 and S240, S244), caveolin-1, collagen VI, annexin I, pGSK3AB (S21 and S9), IGFBP2, and pSTAT3 (Y705) proteins were significantly lower in POSTN-knockdown GSC272 cells than in GSC272 control cells (Fig. 6A). Furthermore, the expression of pS6 (S235, S236 and S240, S244), cyclin B1, TFRC, MEK1, caveolin-1, YB1, and VEGFR2 proteins was markedly lower in POSTN-knockdown GSC272 murine glioma samples than in control glioma samples (Fig. 6A). In both cell lines and tumors from mice, caveolin-1 and pSTAT3 level were decreased by POSTN shRNA. We also examined the protein expression in a murine tumor sample in vivo, expression of caveolin-1, pSTAT3, POSTN, and HIF1α was lower in a POSTN shRNA GSC272–derived mouse tumor than in a parental GSC272-derived mouse tumor (Fig. 6B). Genes identified as regulated by POSTN both in vitro and in vivo were subjected to Ingenuity Pathway Analysis (QIAGEN; Fig. 6C).

This is the first demonstration that knockdown of POSTN decreases resistance to anti-VEGF-A therapy in human GSCs in vivo. POSTN and VEGF-A are strongly associated with GSC progression. A previous study demonstrated that POSTN expression correlated directly with tumor grade and recurrence and inversely with survival at all grades of adult human glioma (17). We showed that POSTN was expressed in GSC6-27, GSC11, and GSC272 cells. Also, mice implanted with GSC11 and GSC272 cells increased POSTN expression by treatment with bevacizumab. Knockdown of POSTN decreased tumor growth and prolonged mouse survival as effectively as treatment with bevacizumab. Mice implanted with GSCs infected with POSTN shRNA and treated with bevacizumab exhibited a synergistic effect on tumor volume and survival.

The TGFβ-induced (TGFBI; BIGH3) protein and POSTN have the same EMI domain and four highly conserved fasciclin I domains. However, they have different C-terminal domains; TGFBI has the RGD motif, whereas POSTN has a more extended C-terminal domain. TGFBI can bind with ECM (33) and POSTN (34). In addition, the C-terminal domain of POSTN binds to the ECM protein to regulate organization of the cellular matrix (35). GSCs have expression of POSTN but not TGFBI (data not shown). Therefore, the C-terminal domain of POSTN may be a key domain for GSC invasion.

In our PCR-based array experiment, POSTN influenced the expression of several angiogenesis- and EMT-related genes in GSC272 cells. Specifically, VEGF-A expression was lower in cells with knockdown of POSTN expression than in control cells. Also, HIF1α activates transcription of various hypoxia-inducible genes, such as VEGF-A, and ECM metabolism (urokinase-type plasminogen activator receptor and MMPs; ref. 36). We found that POSTN affected HIF1α. Furthermore, MMP3 and MMP9 expression was lower in GSC272 cells with POSTN knockdown than in control cells in our PCR array. In addition, EMT plays a critical role in invasiveness of cancer cells, and during EMT, cancer cells regulate expression of mesenchymal genes such as vimentin, fibronectin, MMPs, and N-cadherin. As an ECM protein, POSTN regulates cell adhesion (37). In pancreatic tumor cells, POSTN induces their invasiveness by downregulating E-cadherin expression via Snail (38). In our study, recombinant POSTN increased the invasiveness of GSC272 and GSC11 cells. Knockdown of POSTN expression resulted in less invasion and lower expression of the EMT-related genes N-cadherin and caveolin. It has been reported previously that caveolin-1 expression was downregulated in the cores and penumbrae of ischemic rat brains (39). In another study, caveolin-1 affected VEGF, p42/44 MAPK, PI3K/AKT, and STAT3 signaling in neural progenitor stem cells (40). Finally, the STAT3 pathway is a potent regulator of the hypoxic pathway, and the STAT3 inhibitor AZD1480 may circumvent the induction of hypoxia in glioblastoma (41). More interestingly, POSTN expression can be regulated by STAT3. We observed that treatment with AZD1480 abolished the expression of POSTN as well as HIF1α, VEGF-A, and caveolin-1 in GSC272. Also, recombinant POSTN phosphorylated STAT3. Collectively, these results suggest that POSTN and STAT3 are controlled interactively. Studies elucidating the mechanisms of caveolin-related invasion in glioma are being planned.

We found that TGFβ1 increased the secretion of POSTN, which regulates several EMT and angiogenesis markers in vitro. In a previous study, TGFβ1 induced ECM proteins involved in cell survival, angiogenesis, and invasion (12). In another, POSTN decreased HIF1α accumulation by blocking TGFβ1 type I receptor signaling in human PDL cells in vitro (42). Here we observed that, in GSC cells, POSTN secretion was increased by TGFβ1 and blocked by treatment with an inhibitor of the TGFβ receptor. After treatment with TGFβ1, phosphorylation of SMAD3 and SMAD2 increased in GSC272 cells. POSTN appears to be a key regulator of TGFβ1-induced invasion and VEGF-A expression, and TGFβ1 increases the expression of HIF1α.

The expression of the αvβ3 and αvβ5 integrin heterodimer subtypes correlates directly with glioma grade (43). Ishida and colleagues (44) reported that an antagonist of integrin αvβ3 and integrin αvβ5 prevented bevacizumab-induced invasion in orthotopic glioma models that express these integrins at high levels. We found that GSC11 cells expressed the POSTN receptors integrin β1 and integrin β3 and that GSC272 cells expressed integrin β1 more highly than integrin β3. Exposure to an antibody against integrin β1 blocked invasiveness induced by TGFβ1 and POSTN in GSC11 and GSC272 cells but integrin β3 did not.

Glioblastoma is a highly vascularized tumor, and VEGF-A is a key regulator of tumor angiogenesis. In animal models, bevacizumab inhibits angiogenesis and tumor growth (45). Anti-VEGF therapy alone results in resistance, including increased expression of growth factors, angiopoietin1, VEGF, stromal cell–derived factor-1A, stem cell factors, STAT3, and osteopontin (46–49). Several factors are induced by hypoxia and increased HIF1α protein expression. We know that TGFβ expression in GSC11 cells is increased by bevacizumab-based treatment in mice (10), but the mechanism of its regulation has yet to be determined. In mice implanted with POSTN shRNA-infected GSC272 cells, resistance to bevacizumab-based treatment decreased the expression of both HIF1α and TGFβ1. During tumor growth in mice, POSTN was expressed in GSCs while TGFβ1 was expressed by macrophages. Anti-VEGF-A therapy is known to increase the infiltration of macrophages (50). After inflammation or mechanical stress, expression of TGFβ1 increases in macrophages and neutrophils (37). Our results demonstrate that treatment with bevacizumab increased POSTN expression in GSCs, which mediates TGFβ1 expression in macrophages. Malanchi and colleagues. (15) reported that POSTN functions as a bridge between cancer stem cells and their metastatic niches. Evidence reported herein suggests that POSTN regulates interaction between GSCs and their environment during anti-VEGF-A therapy. Targeting POSTN may represent a novel method to overcome the highly infiltrative phenotype of glioblastoma and prevent resistance to antiangiogenic therapy.

J.F. de Groot is a consultant/advisory board member for Genentech/Roche. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S.Y. Park, Y. Piao, J.F. de Groot

Development of methodology: S.Y. Park

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.Y. Park, J. Dong

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.Y. Park, Y. Piao, K.J. Jeong, J. Dong

Writing, review, and/or revision of the manuscript: S.Y. Park, J.F. de Groot

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Piao, K.J. Jeong

Study supervision: S.Y. Park, J.F. de Groot

The authors thank Kay Hyde and Kathryn Hale for editorial assistance.

This work was supported in part by the M.G. Williams Memorial Brain Tumor Research Fund (to J.F. de Groot) and NCI Cancer Center Support Grant (P30 CA016672; to R.A. DePinho).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Arrillaga-Romany
I
,
Reardon
DA
,
Wen
PY
. 
Current status of antiangiogenic therapies for glioblastomas
.
Expert Opin Investig Drugs
2014
;
23
:
199
210
.
2.
Liu
LX
,
Lu
H
,
Luo
Y
,
Date
T
,
Belanger
AJ
,
Vincent
KA
, et al
Stabilization of vascular endothelial growth factor mRNA by hypoxia-inducible factor 1
.
Biochem Biophys Res Commun
2002
;
291
:
908
14
.
3.
Bao
S
,
Wu
Q
,
Sathornsumetee
S
,
Hao
Y
,
Li
Z
,
Hjelmeland
AB
, et al
Stem cell-like glioma cells promote tumor angiogenesis through vascular endothelial growth factor
.
Cancer Res
2006
;
66
:
7843
8
.
4.
Takano
S
,
Yoshii
Y
,
Kondo
S
,
Suzuki
H
,
Maruno
T
,
Shirai
S
, et al
Concentration of vascular endothelial growth factor in the serum and tumor tissue of brain tumor patients
.
Cancer Res
1996
;
56
:
2185
90
.
5.
Soda
Y
,
Myskiw
C
,
Rommel
A
,
Verma
IM
. 
Mechanisms of neovascularization and resistance to anti-angiogenic therapies in glioblastoma multiforme
.
J Mol Med
2013
;
91
:
439
48
.
6.
Leon
SP
,
Folkerth
RD
,
Black
PM
. 
Microvessel density is a prognostic indicator for patients with astroglial brain tumors
.
Cancer
1996
;
77
:
362
72
.
7.
Phillips
HS
,
Kharbanda
S
,
Chen
R
,
Forrest
WF
,
Soriano
RH
,
Wu
TD
, et al
Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis
.
Cancer Cell
2006
;
9
:
157
73
.
8.
Ferrara
N
,
Hillan
KJ
,
Gerber
HP
,
Novotny
W
. 
Discovery and development of bevacizumab, an anti-VEGF antibody for treating cancer
.
Nat Rev Drug Discov
2004
;
3
:
391
400
.
9.
de Groot
JF
,
Fuller
G
,
Kumar
AJ
,
Piao
Y
,
Eterovic
K
,
Ji
Y
, et al
Tumor invasion after treatment of glioblastoma with bevacizumab: radiographic and pathologic correlation in humans and mice
.
Neuro Oncol
2010
;
12
:
233
42
.
10.
Piao
Y
,
Liang
J
,
Holmes
L
,
Henry
V
,
Sulman
E
,
de Groot
JF
. 
Acquired resistance to anti-VEGF therapy in glioblastoma is associated with a mesenchymal transition
.
Clin Cancer Res
2013
;
19
:
4392
403
.
11.
Norris
RA
,
Potts
JD
,
Yost
MJ
,
Junor
L
,
Brooks
T
,
Tan
H
, et al
Periostin promotes a fibroblastic lineage pathway in atrioventricular valve progenitor cells
.
Dev Dyn
2009
;
238
:
1052
63
.
12.
Kim
JE
,
Kim
SJ
,
Lee
BH
,
Park
RW
,
Kim
KS
,
Kim
IS
. 
Identification of motifs for cell adhesion within the repeated domains of transforming growth factor-beta-induced gene, betaig-h3
.
J Biol Chem
2000
;
275
:
30907
15
.
13.
Politz
O
,
Gratchev
A
,
McCourt
PA
,
Schledzewski
K
,
Guillot
P
,
Johansson
S
, et al
Stabilin-1 and -2 constitute a novel family of fasciclin-like hyaluronan receptor homologues
.
Biochem J
2002
;
362
:
155
64
.
14.
Takeshita
S
,
Kikuno
R
,
Tezuka
K
,
Amann
E
. 
Osteoblast-specific factor 2: cloning of a putative bone adhesion protein with homology with the insect protein fasciclin I
.
Biochem J
1993
;
294
:
271
8
.
15.
Malanchi
I
,
Santamaria-Martinez
A
,
Susanto
E
,
Peng
H
,
Lehr
HA
,
Delaloye
JF
, et al
Interactions between cancer stem cells and their niche govern metastatic colonization
.
Nature
2012
;
481
:
85
9
.
16.
Wang
H
,
Wang
Y
,
Jiang
C
. 
Stromal protein periostin identified as a progression associated and prognostic biomarker in glioma via inducing an invasive and proliferative phenotype
.
Int J Oncol
2013
;
42
:
1716
24
.
17.
Mikheev
AM
,
Mikheeva
SA
,
Trister
AD
,
Tokita
MJ
,
Emerson
SN
,
Parada
CA
, et al
Periostin is a novel therapeutic target that predicts and regulates glioma malignancy
.
Neuro Oncol
2015
;
17
:
372
82
.
18.
Butcher
JT
,
Norris
RA
,
Hoffman
S
,
Mjaatvedt
CH
,
Markwald
RR
. 
Periostin promotes atrioventricular mesenchyme matrix invasion and remodeling mediated by integrin signaling through Rho/PI 3-kinase
.
Dev Biol
2007
;
302
:
256
66
.
19.
Nuzzo
PV
,
Buzzatti
G
,
Ricci
F
,
Rubagotti
A
,
Argellati
F
,
Zinoli
L
, et al
Periostin: a novel prognostic and therapeutic target for genitourinary cancer?
Clin Genitourin Cancer
2014
;
12
:
301
11
.
20.
Yan
W
,
Shao
R
. 
Transduction of a mesenchyme-specific gene periostin into 293T cells induces cell invasive activity through epithelial-mesenchymal transformation
.
J Biol Chem
2006
;
281
:
19700
8
.
21.
Kudo
Y
,
Iizuka
S
,
Yoshida
M
,
Nguyen
PT
,
Siriwardena
SB
,
Tsunematsu
T
, et al
Periostin directly and indirectly promotes tumor lymphangiogenesis of head and neck cancer
.
PLoS One
2012
;
7
:
e44488
.
22.
Horiuchi
K
,
Amizuka
N
,
Takeshita
S
,
Takamatsu
H
,
Katsuura
M
,
Ozawa
H
, et al
Identification and characterization of a novel protein, periostin, with restricted expression to periosteum and periodontal ligament and increased expression by transforming growth factor beta
.
J Bone Miner Res
1999
;
14
:
1239
49
.
23.
Ji
X
,
Chen
D
,
Xu
C
,
Harris
SE
,
Mundy
GR
,
Yoneda
T
. 
Patterns of gene expression associated with BMP-2-induced osteoblast and adipocyte differentiation of mesenchymal progenitor cell 3T3-F442A
.
J Bone Miner Metab
2000
;
18
:
132
9
.
24.
Tada
T
,
Yabu
K
,
Kobayashi
S
. 
Detection of active form of transforming growth factor-beta in cerebrospinal fluid of patients with glioma
.
Jpn J Cancer Res
1993
;
84
:
544
8
.
25.
Leitlein
J
,
Aulwurm
S
,
Waltereit
R
,
Naumann
U
,
Wagenknecht
B
,
Garten
W
, et al
Processing of immunosuppressive pro-TGF-beta 1,2 by human glioblastoma cells involves cytoplasmic and secreted furin-like proteases
.
J Immunol
2001
;
166
:
7238
43
.
26.
Singh
SK
,
Hawkins
C
,
Clarke
ID
,
Squire
JA
,
Bayani
J
,
Hide
T
, et al
Identification of human brain tumour initiating cells
.
Nature
2004
;
432
:
396
401
.
27.
Piao
Y
,
Liang
J
,
Holmes
L
,
Zurita
AJ
,
Henry
V
,
Heymach
JV
, et al
Glioblastoma resistance to anti-VEGF therapy is associated with myeloid cell infiltration, stem cell accumulation, and a mesenchymal phenotype
.
Neuro Oncol
2012
;
14
:
1379
92
.
28.
Lucio-Eterovic
AK
,
Piao
Y
,
de Groot
JF
. 
Mediators of glioblastoma resistance and invasion during antivascular endothelial growth factor therapy
.
Clin Cancer Res
2009
;
15
:
4589
99
.
29.
Kim
K
,
Lu
Z
,
Hay
ED
. 
Direct evidence for a role of beta-catenin/LEF-1 signaling pathway in induction of EMT
.
Cell Biol Int
2002
;
26
:
463
76
.
30.
Davies
M
,
Robinson
M
,
Smith
E
,
Huntley
S
,
Prime
S
,
Paterson
I
. 
Induction of an epithelial to mesenchymal transition in human immortal and malignant keratinocytes by TGF-beta1 involves MAPK, Smad and AP-1 signalling pathways
.
J Cell Biochem
2005
;
95
:
918
31
.
31.
Nawshad
A
,
Lagamba
D
,
Polad
A
,
Hay
ED
. 
Transforming growth factor-beta signaling during epithelial-mesenchymal transformation: implications for embryogenesis and tumor metastasis
.
Cells Tissues Organs
2005
;
179
:
11
23
.
32.
Hardy
KM
,
Booth
BW
,
Hendrix
MJ
,
Salomon
DS
,
Strizzi
L
. 
ErbB/EGF signaling and EMT in mammary development and breast cancer
.
J J Mammary Gland Biol Neoplasia
2010
;
15
:
191
9
.
33.
Hashimoto
K
,
Noshiro
M
,
Ohno
S
,
Kawamoto
T
,
Satakeda
H
,
Akagawa
Y
, et al
Characterization of a cartilage-derived 66-kDa protein (RGD-CAP/beta ig-h3) that binds to collagen
.
B Biochim Biophys Acta
1997
;
1355
:
303
14
.
34.
Kim
BY
,
Olzmann
JA
,
Choi
SI
,
Ahn
SY
,
Kim
TI
,
Cho
HS
, et al
Corneal dystrophy-associated R124H mutation disrupts TGFBI interaction with Periostin and causes mislocalization to the lysosome
.
J Biol Chem
2009
;
284
:
19580
91
.
35.
Hoersch
S
,
Andrade-Navarro
MA
. 
Periostin shows increased evolutionary plasticity in its alternatively spliced region
.
BMC Evol Biol
2010
;
10
:
30
.
36.
Semenza
GL
. 
Targeting HIF-1 for cancer therapy
.
Nat Rev Cancer
2003
;
3
:
721
32
.
37.
Takayama
G
,
Arima
K
,
Kanaji
T
,
Toda
S
,
Tanaka
H
,
Shoji
S
, et al
Periostin: a novel component of subepithelial fibrosis of bronchial asthma downstream of IL-4 and IL-13 signals
.
J Allergy Clin Immunol
2006
;
118
:
98
104
.
38.
Kim
CJ
,
Sakamoto
K
,
Tambe
Y
,
Inoue
H
. 
Opposite regulation of epithelial-to-mesenchymal transition and cell invasiveness by periostin between prostate and bladder cancer cells
.
Int J Oncol
2011
;
38
:
1759
66
.
39.
Shen
J
,
Ma
S
,
Chan
P
,
Lee
W
,
Fung
PC
,
Cheung
RT
, et al
Nitric oxide down-regulates caveolin-1 expression in rat brains during focal cerebral ischemia and reperfusion injury
.
J Neurochem
2006
;
96
:
1078
89
.
40.
Temple
S
. 
The development of neural stem cells
.
Nature
2001
;
414
:
112
7
.
41.
de Groot
J
,
Liang
J
,
Kong
LY
,
Wei
J
,
Piao
Y
,
Fuller
G
, et al
Modulating antiangiogenic resistance by inhibiting the signal transducer and activator of transcription 3 pathway in glioblastoma
.
Oncotarget
2012
;
3
:
1036
48
.
42.
Aukkarasongsup
P
,
Haruyama
N
,
Matsumoto
T
,
Shiga
M
,
Moriyama
K
. 
Periostin inhibits hypoxia-induced apoptosis in human periodontal ligament cells via TGF-beta signaling
.
Biochem Biophys Res Commun
2013
;
441
:
126
32
.
43.
Schnell
O
,
Krebs
B
,
Wagner
E
,
Romagna
A
,
Beer
AJ
,
Grau
SJ
, et al
Expression of integrin alphavbeta3 in gliomas correlates with tumor grade and is not restricted to tumor vasculature
.
Brain Pathol
2008
;
18
:
378
86
.
44.
Ishida
J
,
Onishi
M
,
Kurozumi
K
,
Ichikawa
T
,
Fujii
K
,
Shimazu
Y
, et al
Integrin inhibitor suppresses bevacizumab-induced glioma invasion
.
Transl Oncol
2014
;
7
:
292
302
.
45.
Presta
LG
,
Chen
H
,
O'Connor
SJ
,
Chisholm
V
,
Meng
YG
,
Krummen
L
, et al
Humanization of an anti-vascular endothelial growth factor monoclonal antibody for the therapy of solid tumors and other disorders
.
Cancer Res
1997
;
57
:
4593
9
.
46.
Batchelor
TT
,
Sorensen
AG
,
di Tomaso
E
,
Zhang
WT
,
Duda
DG
,
Cohen
KS
, et al
AZD2171, a pan-VEGF receptor tyrosine kinase inhibitor, normalizes tumor vasculature and alleviates edema in glioblastoma patients
.
Cancer Cell
2007
;
11
:
83
95
.
47.
Casanovas
O
,
Hicklin
DJ
,
Bergers
G
,
Hanahan
D
. 
Drug resistance by evasion of antiangiogenic targeting of VEGF signaling in late-stage pancreatic islet tumors
.
Cancer Cell
2005
;
8
:
299
309
.
48.
Fernando
NT
,
Koch
M
,
Rothrock
C
,
Gollogly
LK
,
D'Amore
PA
,
Ryeom
S
, et al
Tumor escape from endogenous, extracellular matrix-associated angiogenesis inhibitors by up-regulation of multiple proangiogenic factors
.
Clin Cancer Res
2008
;
14
:
1529
39
.
49.
Ebos
JM
,
Lee
CR
,
Kerbel
RS
. 
Tumor and host-mediated pathways of resistance and disease progression in response to antiangiogenic therapy
.
Clin Cancer Res
2009
;
15
:
5020
5
.
50.
Liang
J
,
Piao
Y
,
Holmes
L
,
Fuller
GN
,
Henry
V
,
Tiao
N
, et al
Neutrophils promote the malignant glioma phenotype through S100A4
.
Clin Cancer Res
2014
;
20
:
187
98
.