Portal de Periódicos da CAPES

Endothelium-Independent Primitive Myxoid Vascularization Creates Invertebrate-Like Channels to Maintain Blood Supply in Optic Gliomas

2017; Elsevier BV; Volume: 187; Issue: 8 Linguagem: Inglês

10.1016/j.ajpath.2017.04.004

1525-2191

Matija Snuderl, Guoan Zhang, Pamela Wu, Tara Jennings, Seema Shroff, Valerio Ortenzi, Rajan Jain, Benjamin Cohen, Jason Reidy, Mitchell S. Dushay, Jeffrey H. Wisoff, David H. Harter, Matthias A. Karajannis, David Fenyö, Thomas A. Neubert, David Zagzag,

Retinal Diseases and Treatments

Optic gliomas are brain tumors characterized by slow growth, progressive loss of vision, and limited therapeutic options. Optic gliomas contain various amounts of myxoid matrix, which can represent most of the tumor mass. We sought to investigate biological function and protein structure of the myxoid matrix in optic gliomas to identify novel therapeutic targets. We reviewed histological features and clinical imaging properties, analyzed vasculature by immunohistochemistry and electron microscopy, and performed liquid chromatography–mass spectrometry on optic gliomas, which varied in the amount of myxoid matrix. We found that although subtypes of optic gliomas are indistinguishable on imaging, the microvascular network of pilomyxoid astrocytoma, a subtype of optic glioma with abundant myxoid matrix, is characterized by the presence of endothelium-free channels in the myxoid matrix. These tumors show normal perfusion by clinical imaging and lack histological evidence of hemorrhage organization or thrombosis. The myxoid matrix is composed predominantly of the proteoglycan versican and its linking protein, a vertebrate hyaluronan and proteoglycan link protein 1. We propose that pediatric optic gliomas can maintain blood supply without endothelial cells by using invertebrate-like channels, which we termed primitive myxoid vascularization. Enzymatic targeting of the proteoglycan versican/hyaluronan and proteoglycan link protein 1 rich myxoid matrix, which is in direct contact with circulating blood, can provide novel therapeutic avenues for optic gliomas of childhood. Optic gliomas are brain tumors characterized by slow growth, progressive loss of vision, and limited therapeutic options. Optic gliomas contain various amounts of myxoid matrix, which can represent most of the tumor mass. We sought to investigate biological function and protein structure of the myxoid matrix in optic gliomas to identify novel therapeutic targets. We reviewed histological features and clinical imaging properties, analyzed vasculature by immunohistochemistry and electron microscopy, and performed liquid chromatography–mass spectrometry on optic gliomas, which varied in the amount of myxoid matrix. We found that although subtypes of optic gliomas are indistinguishable on imaging, the microvascular network of pilomyxoid astrocytoma, a subtype of optic glioma with abundant myxoid matrix, is characterized by the presence of endothelium-free channels in the myxoid matrix. These tumors show normal perfusion by clinical imaging and lack histological evidence of hemorrhage organization or thrombosis. The myxoid matrix is composed predominantly of the proteoglycan versican and its linking protein, a vertebrate hyaluronan and proteoglycan link protein 1. We propose that pediatric optic gliomas can maintain blood supply without endothelial cells by using invertebrate-like channels, which we termed primitive myxoid vascularization. Enzymatic targeting of the proteoglycan versican/hyaluronan and proteoglycan link protein 1 rich myxoid matrix, which is in direct contact with circulating blood, can provide novel therapeutic avenues for optic gliomas of childhood. Optic gliomas are low-grade astrocytic tumors of the optic pathway and occur predominantly in children and young adults. The factors driving the behavior of optic gliomas remain unknown. The genetic landscape of this tumor subtype is relatively homogeneous and insufficient to explain the diverse phenotypes and unpredictable clinical courses observed.1Rodriguez F.J. Ligon A.H. Horkayne-Szakaly I. Rushing E.J. Ligon K.L. Vena N. Garcia D.I. Cameron J.D. Eberhart C.G. BRAF duplications and MAPK pathway activation are frequent in gliomas of the optic nerve proper.J Neuropathol Exp Neurol. 2012; 71: 789-794Crossref PubMed Scopus (51) Google Scholar, 2Rodriguez F.J. Lim K.S. Bowers D. Eberhart C.G. Pathological and molecular advances in pediatric low-grade astrocytoma.Annu Rev Pathol. 2013; 8: 361-379Crossref PubMed Scopus (63) Google Scholar It is possible that the tumor microenvironment plays an important role in the development of optic gliomas, but in contrast to malignant gliomas, angiogenesis is not associated with poor outcomes. Optic gliomas typically grow slowly, and may lead to loss of vision, endocrine deficiencies, and neurological impairment. The role of surgery is limited because of the critical location and infiltrative growth pattern of optic gliomas. Other treatment modalities, including chemotherapy and radiation therapy, carry significant morbidity and often fail to provide tumor control. Optic gliomas are composed of two distinct histological subtypes, pilocytic astrocytoma (PA) and pilomyxoid astrocytoma (PMXA).3Louis D.N. Ohgaki H. Wiestler O.D. Cavenee W.K. Ellison D.W. Figarella-Branger D. Perry A. Reifenberger G. von Deimling A. WHO Classification of Tumors of the Central Nervous System. IARC Press, Lyon2016Google Scholar Abundant bluish chondroid myxoid matrix, after hematoxylin and eosin (H&E) staining, is characteristic of PMXA but not PA, and typically accounts for >50% of tumor volume. The biological role of the matrix protein composition or its effect on tumor growth and angiogenesis in PMXA is currently unknown. Interestingly, PAs are commonly diagnosed throughout the brain, whereas PMXAs arise almost exclusively in the optic tract, suggesting a role of optic tract microenvironment in the development of this particular subtype of low-grade gliomas. The prognostic significance of the myxoid matrix remains unknown, partially because of a large number of tumors with intermediate features containing some myxoid matrix but not fulfilling all the criteria for diagnosis of PMXA. The absence of clear prognostic value of pilomyxoid features recently led to changing the grade of PMXA from World Health Organization (WHO) grade II, and the current recommendation is not to assign a WHO grade.3Louis D.N. Ohgaki H. Wiestler O.D. Cavenee W.K. Ellison D.W. Figarella-Branger D. Perry A. Reifenberger G. von Deimling A. WHO Classification of Tumors of the Central Nervous System. IARC Press, Lyon2016Google Scholar Endothelial cells are necessary for angiogenesis in vertebrates, but not in invertebrates.4Munoz-Chapuli R. Evolution of angiogenesis.Int J Dev Biol. 2011; 55: 345-351Crossref PubMed Scopus (43) Google Scholar Hypoxia is a strong stimulator of angiogenesis in tumors and is driven largely by vascular endothelial growth factor A secretion. Vascular endothelial growth factor A mediates vasculogenesis by recruiting endothelial progenitors from circulating bone marrow–derived cells.5Hardee M.E. Zagzag D. Mechanisms of glioma-associated neovascularization.Am J Pathol. 2012; 181: 1126-1141Abstract Full Text Full Text PDF PubMed Scopus (313) Google Scholar, 6Carmeliet P. Jain R.K. Molecular mechanisms and clinical applications of angiogenesis.Nature. 2011; 473: 298-307Crossref PubMed Scopus (3719) Google Scholar Vascular endothelial growth factor A–driven vascular proliferation is a hallmark of malignant gliomas and is associated with aggressive tumor behavior and poor survival.3Louis D.N. Ohgaki H. Wiestler O.D. Cavenee W.K. Ellison D.W. Figarella-Branger D. Perry A. Reifenberger G. von Deimling A. WHO Classification of Tumors of the Central Nervous System. IARC Press, Lyon2016Google Scholar During evolution, three major types of circulatory system developed in metazoans: hemal, hemocoelic, and endothelial system. Although vertebrates have a closed endothelial circulatory system and endothelial cells represent a key element of vascular anatomy, the hemal and hemocoelic systems of invertebrates lack endothelial cells. The hemal system of invertebrates consists of a network of spaces lined by basement membranes internally and endodermal or mesodermal epithelial lining on the opposite side.7Munoz-Chapuli R. Carmona R. Guadix J.A. Macias D. Perez-Pomares J.M. The origin of the endothelial cells: an evo-devo approach for the invertebrate/vertebrate transition of the circulatory system.Evol Dev. 2005; 7: 351-358Crossref PubMed Scopus (80) Google Scholar, 8Shigei T. Tsuru H. Ishikawa N. Yoshioka K. Absence of endothelium in invertebrate blood vessels: significance of endothelium and sympathetic nerve/medial smooth muscle in the vertebrate vascular system.Jpn J Pharmacol. 2001; 87: 253-260Crossref PubMed Scopus (30) Google Scholar In the hemocoelic system, the hemal system opens to the coelomic cavity and blood freely circulates around the organs, which is frequently associated with disappearance of the coelomic epithelium.4Munoz-Chapuli R. Evolution of angiogenesis.Int J Dev Biol. 2011; 55: 345-351Crossref PubMed Scopus (43) Google Scholar, 7Munoz-Chapuli R. Carmona R. Guadix J.A. Macias D. Perez-Pomares J.M. The origin of the endothelial cells: an evo-devo approach for the invertebrate/vertebrate transition of the circulatory system.Evol Dev. 2005; 7: 351-358Crossref PubMed Scopus (80) Google Scholar The question of whether human tumors can develop alternative, invertebrate-like, mechanisms for obtaining blood supply led us to discover a biologically novel endothelium-independent mechanism of tumor vascularization distinct from the mechanism of conventional vertebrate angiogenesis. Currently, there are no non-Nf1 murine models of optic glioma or pilomyxoid astrocytoma. Because of the inability of genomic analysis to explain the differences between pilocytic and pilomyxoid astrocytomas, we chose a proteomic-based approach to identify both structure and biology of the myxoid matrix. Herein, we show that pediatric optic pathway gliomas maintain their blood supply and neoplastic cell growth independent of endothelialized blood vessels. Using high-throughput proteomic analysis, we describe, for the first time, the protein structure of the myxoid matrix in optic gliomas and identify potential extracellular matrix targets for therapy of this disease. Clinical information was obtained from the medical records on 120 patients diagnosed as having optic glioma at New York University between 1996 and 2014 after obtaining Institutional Review Board approval. Axial T2, precontrast, and post-contrast T1 weighted magnetic resonance images were reviewed by two investigators (R.J. and B.C.). Diameters, margins, heterogeneity, and percentage of the contrast enhancement, extent of edema, and presence or absence of susceptibility, cysts, necrosis, and hemorrhage were evaluated. Perfusion was evaluated where available. Because of the changes in imaging techniques, a complete set of comparable clinical imaging studies was available for 10 optic gliomas. Histological features of the samples were evaluated by two neuropathologists (M.S. and D.Z.) using standard, clinical H&E staining. Following clinically approved protocols, the following antibodies were used for immunohistochemical analysis of 12 optic gliomas for which sufficient tissue was available: CD34, prediluted (catalog number 790-2927; Ventana Medical Systems, Tuscon, AZ); Erg, prediluted (catalog number 790-4576; Ventana Medical Systems); GLUT-1, prediluted (catalog number 355A-18; Cell Marque, Rocklin CA); CA-IX, prediluted (catalog number 379M-18; Cell Marque); and Ki-67, prediluted (catalog number 790-4286; Ventana Medical Systems). Microvascular density was assessed as described previously9di Tomaso E. Snuderl M. Kamoun W.S. Duda D.G. Auluck P.K. Fazlollahi L. Andronesi O.C. Frosch M.P. Wen P.Y. Plotkin S.R. Hedley-Whyte E.T. Sorensen A.G. Batchelor T.T. Jain R.K. Glioblastoma recurrence after cediranib therapy in patients: lack of "rebound" revascularization as mode of escape.Cancer Res. 2011; 71: 19-28Crossref PubMed Scopus (158) Google Scholar and presented as number of blood vessels per mm2. The proliferation rate was calculated as a ration of Ki-67–positive cells after 500 total tumor cells were counted. Data are expressed as means ± SEM. The principal statistical test was the t-test (two-tailed with unequal variance). We considered a value of P < 0.05 to be statistically significant. The Versican isoform V1 expression was confirmed using anti-Versican antibody (ab19345) from Abcam (Cambridge, MA) after heat-induced antigen retrieval (dilution, 1:100). Three optic pilomyxoid astrocytomas had tissue collected at the time of surgery for electron microscopy studies. Specimens for electron microscopy were fixed in phosphate-buffered 1% glutaraldehyde/4% paraformaldehyde, post-fixed in 1% osmium teroxide, and embedded in epoxy resin in a standard manner. Scout sections (1 μm thick) were stained with toluidine blue; thin (90 nm thick) sections were collected on open-slot collodion-coated grids, stained with uranyl acetate and lead citrate, and examined in a Zeiss EM900 electron microscope (ZEISS, Oberkochen, Germany) at 80 kilovolts. Sections were independently evaluated by two observers (J.J.R. and M.S.). Liquid chromatography–mass spectrometry (LC-MS) was performed on 10 snap-frozen tumors previously collected by the New York University Brain Tissue Repository. Tissue samples were lysed by sonication in a solution containing 9 mol/L urea, 20 mmol/L Tris, pH 8, 0.2 mmol/L EDTA, and protease inhibitors (Complete tablet; Roche, Mannheim, Germany). Proteins were then reduced with dithiothreitol and alkylated with iodoacetamide before overnight digestion with trypsin at 37°C. The tryptic peptides were desalted using StageTips (Thermo Fischer Scientific, West Palm Beach, FL) and dried. Each biological replicate was analyzed three times by LC–MS, using a Thermo Scientific EASY-nLC 1000 coupled to a Q Exactive mass spectrometer (Thermo Fisher Scientific). A self-packed 75-μm × 25-cm reversed-phase column (Reprosil C18; 3 μm; Dr. Maisch HPLC GmbH, Ammerbuch, Germany) was used for peptide separation. Peptides were eluted by a gradient of 3% to 30% acetonitrile in 0.1% formic acid over 180 minutes at a flow rate of 250 nL/minute at 45°C. The Q Exactive was operated in data-dependent mode with survey scans acquired at a resolution of 50,000 at m/z 400 (transient time, 256 milliseconds). Up to the top 10 most abundant precursors from the survey scan were selected with an isolation window of 1.6 Thomsons and fragmented by higher-energy collisional dissociation with normalized collision energies of 27. The maximum ion injection times for the survey scan and the MS/MS scans were 20 and 60 milliseconds, respectively, and the ion target value for both scan modes was set to 1,000,000. The raw files were processed using the MaxQuant computational proteomics platform version 1.2.7.0 (Max Planck Institute, Munich, Germany) for protein identification. The fragmentation spectra were used to search the UniProt human protein database (downloaded February 8, 2013) containing 87,647 protein sequences and allowing up to two missed tryptic cleavages. Carbamidomethylation of cysteine was set as a fixed modification, and oxidation of methionine and protein N-terminal acetylation were used as variable modifications for database searching. The precursor and fragment mass tolerances were set to 7 and 20 ppm, respectively. Both peptide and protein identifications were filtered at 1% false discovery rate based on decoy search using a database with the protein sequences reversed. For each condition, there were three to four biological replicates and three technical replicates. The R package limma (www.bioconductor.org) was used to perform the following analysis. Raw spectral counts were normalized by the total number of counts per sample and then log2 transformed. Precision weights were calculated for each observation using the voom algorithm, and technical replicates were blocked. A linear model was fit to the data that took into account the interreplicate correlation, the precision weights, and the replicate blocking. limma uses the empirical Bayes method of readjusting the residual variances to increase statistical power for all parallel t-tests. Statistical significance is presented as a q value (ie, a P value that has been adjusted for multiple hypothesis testing). All genes with all positive fold changes and a q value of <0.05 were considered. limma is an open-source software package originally developed for microarray analysis and then later modified to handle RNA-sequencing counts data. It has been shown that the analogous nature of RNA-sequencing counts data and proteomics spectral counts data means that proteomics data sets can take advantage of the statistical tools developed for RNA-sequencing counts data.10Anders S. McCarthy D.J. Chen Y. Okoniewski M. Smyth G.K. Huber W. Robinson M.D. Count-based differential expression analysis of RNA sequencing data using R and Bioconductor.Nat Protoc. 2013; 8: 1765-1786Crossref PubMed Scopus (759) Google Scholar Four databases were used to mine protein-protein interactions (PPIs): BioGRID, SPIKE, IntAct, and APID. IntAct also contains data from MINT. These databases were chosen for having purely evidence-based PPIs. REST APIs developed by PSICQUIC were queried using an in-house script (https://github.com/pambot/GEPPI, last accessed January 17, 2017) to form a network graph file (GLM format) of all PPIs between significantly regulated proteins, which was then imported into Cytoscape for visualization. To explore the role of the myxoid matrix and its contribution to optic glioma biology, we reviewed clinical and pathological data on a cohort of 120 patients with optic gliomas diagnosed at New York University Langone Medical Center from 1996 to 2014. By histology, 43 (36%) of the tumors were classified as PA (Figure 1, E and F) and 25 cases (21%) fit the WHO criteria for a PMXA (Figure 1, C and D).3Louis D.N. Ohgaki H. Wiestler O.D. Cavenee W.K. Ellison D.W. Figarella-Branger D. Perry A. Reifenberger G. von Deimling A. WHO Classification of Tumors of the Central Nervous System. IARC Press, Lyon2016Google Scholar However, 52 cases (43%) could not be definitively classified as either PA or PMXA and were best classified as pilocytic astrocytoma with pilomyxoid features (PA/PMXA). PA/PMXA tumors showed various amounts of myxoid matrix present in the tumor. Furthermore, there was marked intratumoral heterogeneity, with some areas of a tumor showing morphological features of a PA and some of a PMXA. Because optic gliomas are rarely excised completely, there is a possibility of sampling bias that can influence the final histological diagnosis. Therefore, we sought to establish whether PA and PMXA can be distinguished by imaging. When comparing imaging properties of optic gliomas, for which complete comparable preoperative magnetic resonance imaging was available, we found that PA and PMXA are indistinguishable by imaging features (n = 5 per group). All optic gliomas had relatively well-defined margins and showed heterogeneous contrast enhancement, suggesting rich, but abnormally leaky, vasculature. Tumor necrosis was absent in all samples, indicating that neither hypoperfusion nor hypoxia was present (Figure 1, A and B). To characterize the microvascular density of optic PA and PMXA on the tissue level, we performed histological and immunohistochemical analysis of six PAs and six PMXAs. We excluded all tumors that had mixed features (PA/PMXA) because of the high variability between different tissue blocks. Although our imaging analysis suggested equal blood supply and no hypoxia in both PAs and PMXAs, our histological analysis (Figure 1, C–F) and immunohistochemistry (Figure 1, G–N) demonstrated that PMXA tumors, in contrast to PAs, exhibited significantly lower microvascular density. This was defined by the presence of CD34-positive (8.1 versus 14.5 blood vessels/mm2; P < 0.002) (Figures 1, G and I, and 2A ) and Erg-positive (7 versus 13.6 blood vessels/mm2; P = 0.003) (Figures 1, H and J, and 2B) endothelium-lined channels. However, when we evaluated the number of perfused blood vessels using GLUT-1, which highlights erythrocytes as well as brain endothelial cells, there was no difference in the number of perfused blood channels between PMXA and PA (15.4 versus 14.5 blood vessels/mm2) (Figures 1, K and M, and 2C). We defined a perfused blood vessel as a vascular channel filled with GLUT-1–positive erythrocytes, irrespective of the presence of the endothelium. Furthermore, PMXA and PA tumor subtypes demonstrated an absence of both necrosis on H&E staining and hypoxia by immunochemistry (Figure 1, L and N). In addition, both tumors displayed similar tumor cell proliferation rates by Ki-67 (Figure 2D). Together, these findings suggest equal functional perfusion of both tumor subtypes. Last, GLUT-1 immunohistochemistry (Figure 1K) highlighted large perfused channels within the myxoid matrix apparent on H&E staining (Figure 1, C and D). These channels maintained the shapes of blood vessels; however, they were completely devoid of endothelial cells (Figure 1, G and H). None of these channels showed histological signs of thrombus organization, fibrin thrombi, or hemosiderin deposition, suggesting that these were not recent or remote hemorrhages into a myxoid matrix. These channels were completely absent in all PAs, which in all cases showed endothelialized blood vessels (Figure 1, E and F). There are currently no patient-derived or genetically engineered murine models of PMXA, which precludes in vivo analysis of perfusion. However, in three PMXA patients, a tumor perfusion analysis was available, showing that areas of enhancement corresponded to increased blood volume, suggesting preserved perfusion in PMXA (Supplemental Figure S1), which is concordant with previous optic glioma imaging studies.11Vinhais S. Optic gliomas in children: a review on MR imaging findings.European Congress of Radiology. 2015; (Poster ECR 2015/C-1676)Google Scholar, 12Lémery-Magnin M. Perfusion weighted imaging in pediatric low grade glioma.European Congress of Radiology. 2015; (Poster ECR2014/C-1270)Google Scholar Serial sectioning of PMXAs showed that erythrocytes are present in endothelialized blood vessels, which open into spaces surrounded by tumor astrocytes and finally only surrounded by the myxoid matrix (Figure 3). To confirm the absence of endothelial cells in these myxoid channels at the ultrastructural level, we performed electron microscopy and mapped the tumoral vascular network of three PMXAs for which tissue of sufficient quality was available for electron microscopy. We found that PMXAs have regular vertebrate blood vessels with endothelial lining (Figures 2E and 3) as well as channels completely lacking endothelial and smooth muscle cells (Figures 2G and 3). We observed neither myxoid channels partially lined by endothelial cells nor myxoid channels with isolated endothelial cells, confirming complete absence of endothelial cells within the myxoid matrix. Complete lack of endothelial cells by ultrastructural analysis also excludes a possibility of primitive endothelial cells or endothelial cell precursors that would lack CD34 or Erg expression. In addition, we identified a transitional zone between both types of blood channels, which was lined by glial cells in direct contact with erythrocytes (Figures 2F and 3). Channels in the myxoid matrix also lacked any ultrastructural evidence of basement membrane (Figure 2H), arguing against the secondary involution of the endothelial cell layer. The myxoid matrix contained fibrin fibers, which outlined the channel, generating a pseudomembrane in response to erythrocyte pressure on the matrix. These fibrin fibers were present only in the myxoid matrix, but not around normal blood vessels, intermediate channels, or tumor cells (Figure 2, E–G). Therefore, we concluded that fibrin in the myxoid matrix may have been extravasated from the blood because of the lack of an endothelial barrier. Fibrin extravasation from plasma into myxoid matrix may be a possible mechanism by which blood in myxoid channels does not develop fibrin microthrombi. When endothelial cells are not present, blood will coagulate instantly when in contact with tumor cells.13Shoji M. Hancock W.W. Abe K. Micko C. Casper K.A. Baine R.M. Wilcox J.N. Danave I. Dillehay D.L. Matthews E. Contrino J. Morrissey J.H. Gordon S. Edgington T.S. Kudryk B. Kreutzer D.L. Rickles F.R. Activation of coagulation and angiogenesis in cancer: immunohistochemical localization in situ of clotting proteins and vascular endothelial growth factor in human cancer.Am J Pathol. 1998; 152: 399-411PubMed Google Scholar However, myxoid matrix secreted by glioma cells does not trigger coagulation. To identify the biochemical composition of the myxoid matrix, we performed LC-MS without sample fractionation,14Michalski A. Damoc E. Hauschild J.P. Lange O. Wieghaus A. Makarov A. Nagaraj N. Cox J. Mann M. Horning S. Mass spectrometry-based proteomics using Q Exactive, a high-performance benchtop quadrupole Orbitrap mass spectrometer.Mol Cell Proteomics. 2011; 10 (M111.011015)Crossref PubMed Scopus (626) Google Scholar, 15Cox J. Mann M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification.Nat Biotechnol. 2008; 26: 1367-1372Crossref PubMed Scopus (9145) Google Scholar and for quantification, we used peptide spectral counts10Anders S. McCarthy D.J. Chen Y. Okoniewski M. Smyth G.K. Huber W. Robinson M.D. Count-based differential expression analysis of RNA sequencing data using R and Bioconductor.Nat Protoc. 2013; 8: 1765-1786Crossref PubMed Scopus (759) Google Scholar (Figure 4A). Proteomic analysis is superior to immunohistochemistry as it allows unbiased quantitative detection of peptides and proteins. We compared 10 optic glioma samples from non-neurofibromatosis type 1 patients in which sufficient frozen tumor tissue was available for analysis. Patients were sex, age, and BRAF status matched, but differed in final histological diagnosis: PA (n = 4), PMXA (n = 3), and tumors with intermediate pathological features (PA/PMXA, n = 3) (Figure 4, A and B). Clinical data are summarized in Figure 4B. The three groups of tumors differed in the amount of the myxoid matrix present on H&E staining (Figure 4C). In addition, their proteomic profiles separated them into three distinct subgroups by unsupervised analyses matching their histological classification. In total, we identified 5389 proteins, of which 188 were differentially expressed in the three groups (P < 0.05, Benjamini-Hochberg adjustment). Between PA and PMXA, we found that most of differentially expressed proteins (146/188) displayed a positive fold change (defined as increasing in PMXA relative to PA), and a minority (42/188) showed a negative fold change. The most abundant extracellular matrix protein with highest spectral count change between PA and PMXA and highest MS signal intensity and count of matched MS/MS spectra was a large chondroitin sulfate proteoglycan, versican (VCAN). VCAN is a versatile proteoglycan with numerous roles in vascular and tumor extracellular matrix biology.16Wight T.N. Kinsella M.G. Evanko S.P. Potter-Perigo S. Merrilees M.J. Versican and the regulation of cell phenotype in disease.Biochim Biophys Acta. 2014; 1840: 2441-2451Crossref PubMed Scopus (89) Google Scholar, 17Wu Y.J. La Pierre D.P. Wu J. Yee A.J. Yang B.B. The interaction of versican with its binding partners.Cell Res. 2005; 15: 483-494Crossref PubMed Scopus (294) Google Scholar, 18Sheng W. Wang G. Wang Y. Liang J. Wen J. Zheng P.S. Wu Y. Lee V. Slingerland J. Dumont D. Yang B.B. The roles of versican V1 and V2 isoforms in cell proliferation and apoptosis.Mol Biol Cell. 2005; 16: 1330-1340Crossref PubMed Scopus (132) Google Scholar VCAN was identified as the most abundant protein based on total intensity and showed a 3.7-fold increase (Q = 0.000463). The versican molecular exists in at least four different isoforms generated by alternative splicing (V0, V1, V2, and V3). These isoforms differ in the size of the core protein and the number of attached chondroitin sulfate chains and cannot be reliably distinguished by LC-MS. Furthermore, these isoforms are present in different physiological and pathological states. To identify the specific VCAN isoform and confirm in the tissue sections that VCAN is produced directly by tumor cells, we performed immunohistochemistry. Immunohistochemical analysis confirmed VCAN isoform V116Wight T.N. Kinsella M.G. Evanko S.P. Potter-Perigo S. Merrilees M.J. Versican and the regulation of cell phenotype in disease.Biochim Biophys Acta. 2014; 1840: 2441-2451Crossref PubMed Scopus (89) Google Scholar is produced by glioma cells and secreted in the extracellular matrix (Figure 5). VCAN isoform V1 is expressed in brain as well as in some brain tumors. Expression of VCAN is critical for ocular development, and genetic defects in VCAN can cause autosomal dominant vitreoretinal degeneration (Wagner syndrome; Mendelian Inheritance in Man 143200) characterized by an empty vitreous cavity.19Kloeckener-Gruissem B. Bartholdi D. Abdou M.T. Zimmermann D.R. Berger W. Identification of the genetic defect in the original Wagner syndrome family.Mol Vis. 2006; 12: 350-355PubMed Google Scholar, 20Miyamoto T. Inoue H. Sakamoto Y. Kudo E. Naito T. Mikawa T. Mikawa Y. Isashiki Y. Osabe D. Shinohara S. Shiota H. Itakura M. Identification of a novel splice site mutation of the CSPG2 gene in a Japanese family with Wagner syndrome.Invest Ophthalmol Vis Sci. 2005; 46: 2726-2735Crossref PubMed Scopus (66) Google Scholar, 21Mukhopadhyay A. Nikopoulos K. Maugeri A. de Brouwer A.P. van Nouhuys C.E. Boon C.J. Perveen R. Zegers H.A. Wittebol-Post D. van den Biesen P.R. van der Velde-Visser S.D. Brunner H.G. Black G.C. Hoyng C.B. Cremers F.P. Erosive vitreoretinopathy and wagner disease are caused by intronic mutations in CSPG2/Versican that result in an imbalance of splice variants.Invest Ophthalmol Vis Sci. 2006; 47: 3565-3572Crossref P