As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

emmm201302827-sm-0001-SuppFig-S1.pdfPDF document9345K

Figure S1. (A) SNAI1, SNAI2 and E-cadherin are expressed in hGBM L0, but not in the other two lines. (B) ZEB1 is the only EMT-associated factor analyzed that affects outcome in the TCGA dataset (REF). (C) ZEB1 is increased at the tumor edge of hGBM L1. Scale bars 10 µm. (D) Tumor mass cells (asterisks) are characterized by membrane-associated beta-catenin, which is absent in invasive cells. Note that invasive cells show no nuclear accumulation of beta-catenin (arrowheads). Scale bars 10 µm.

Figure S2. (A) Beta-catenin is associated with the cell membrane in both control cells (shGFP) and ZEB1 knockdown, while only controls are positive for ZEB1. In addition, control cells appear more dispersed after plating. Scale bars 10 µm. (B) ZEB1 knockdown reduces ZEB1 mRNA levels in tumor-sphere cultures (representative data from hGBM L1, n = 3, t-test). ZEB1 knockdown increases cellular proliferation approximately twofold, whereas forced expression decreases proliferation rates (hGBM L0, n = 3, one-way ANOVA). (C) Threshold images generated from immunofluorescence staining depicted in Figure 2B. These images are used for measuring invasion indices. (D) Putative miR-200c binding sites in the 5′-UTR of c-MYB, SOX2, OLIG2 and CD133.

Figure S3. (A) ROBO1 is preferentially localized at the cell membrane of control cells (shGFP), and absent in ZEB1 knockdown cells. Scale bar, 10µm. (B) ROBO1 expression is increased by antagonizing miR-200 in ZEB1 knockdown cells, and decreased by forced expression of miR-200 in ZEB1 overexpressing cells. (C) Antagonizing miR-200 increases cell migration (n = 3; Mann–Whitney U-test). (D) ROBO1 knockdown reduces migration in a scratch assay (n = 3, Mann–Whitney U-test). (E) ROBO1 knockdown can be rescued by a non-targeted ROBO1 construct. (F) ROBO1 knockdown reduces tumor invasion in vivo (n = 5 animals each, one-way ANOVA). (G) Forced expression of ROBO1 rescues cell migration in ZEB1 knockdown cells, while ROBO1 knockdown reduces migration in ZEB1 overexpressing cells (n = 3; Mann–Whitney U-test).

Figure S4. (A) Antagonizing miR-200 increases expression of c-MYB and MGMT, as well as chemoresistance in ZEB1 knockdown cells (n=8, one-way ANOVA). (B) Forced expression of miR-200 reduces expression of c-MYB and MGMT in ZEB1 overexpressing cells, and results in reduced chemoresistance (n = 8, one-way ANOVA). (C) Knockdown of c-MYB reduces expression of MGMT, while overexpression of c-MYB increases MGMT levels. Both have no influence on ZEB1 expression. Resistance to TMZ is significantly reduced by c-MYB knockdown in vitro (n = 8, oneway ANOVA). (D) Chemoresistance in ZEB1-knockdown cells can be rescued by expression of c-MYB (n = 8, one-way ANOVA). (E) Knockdown of c-MYB reduces expression of MGMT and chemoresistance in ZEB1-overexpressing cells. (F) Rescue of MYB, MGMT and ZEB1 knockdown with respective untargeted constructs.

Figure S5. (A) Bisulfite genomic sequencing of the MGMT promoter demonstrates no significant methylation changes in ZEB1 knockdown cells. The MGMT promoter was amplified from −555 to +120 relative to the TSS. The frequency of methylation at each CG site was calculated from at least 10 cloned and sequenced molecules for each condition. Data is presented as the average methylation frequency and the position of each CG site relative to the TSS is indicated. (B) ZEB1 knockdown increases survival of tumor-bearing animals after one cycle (five injections) of TMZ treatment in vivo (log-rank test). Expression of c-MYB in ZEB1 knockdown cells abolished this survival benefit (log-rank test).

Figure S6. (A) ZEB1 negative glioblastoma specimens do not stain for ZEB1 at the tumor invasion front. Scale bar 20µm (applies to all images). (B) ZEB1, E-cadherin, SNAI1, SNAI2, and Twist-1 are not expressed in glioblastoma specimens. (C) ZEB1 and the proliferation marker Ki-67 are mutually exclusive in tumor specimens (3.66 ± 0.21% co-localization; n = 14; Venn diagrams illustrate overlapping populations of ZEB1 and Ki-67, and ZEB1 and MGMT, respectively). (D) Neither age, nor KPS scores (post-OP or last recorded) correlate with ZEB1, but a significant correlation was found for the ratio of last recorded to post-OP KPS score. (E) PCNA and EGFR are significantly enriched in ZEB1 positive tumor specimens, while loss of NF1 is significantly more frequent in ZEB1 negative samples (Fisher's exact test). (F) Survival analysis for ZEB1 class by molecular subgroups revealed no significant outcome benefit in either subgroup, but shows a general trend for improved outcome in ZEB1 negative tumors. Note the shorter overall survival in the proliferative subgroup (which has the highest percentage of ZEB1+ specimens). (G) TCGA analysis of the ZEB1/miR-200c/c-MYB pathway shows significant changes in patient outcome. Both MGMT and ROBO1 frequently co-occur with ZEB1 in TCGA samples (p-values are from Fisher's exact test). (H) ZEB1 positive samples show higher levels of p-EGFR in RPPA.

Table S1. Clinical information of patient samples used in this study.

Table S2. Vendor information for antibodies used in this study.

emmm201302827-SourceData-Fig1.pdfPDF document2605KSource Data for Figure 1
emmm201302827-SourceData-Fig2.pdfPDF document2605KSource Data for Figure 2
emmm201302827-SourceData-Fig3.pdfPDF document2605KSource Data for Figure 3
emmm201302827-SourceData-Fig4.pdfPDF document2084KSource Data for Figure 4
emmm201302827-SourceData-Fig5.pdfPDF document1297KSource Data for Figure 5
emmm201302827-SourceData-Fig6.pdfPDF document3042KSource Data for Figure 6
emmm201302827.reviewer_comments.pdfPDF document342KReview Process File

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.