Journal of Clinical Investigation EGFR Phosphorylation of DCBLD2

Supplemental Data

Journal: Journal of Clinical Investigation

EGFR Phosphorylation of DCBLD2 Recruits TRAF6 and Stimulates

Akt-promoted Tumorigenesis

Haizhong Feng, Giselle Y. Lopez, Chung Kwon Kim, Angel Alvarez, Christopher G. Duncan, Ryo Nishikawa, Motoo Nagane, An-Jey Su, Philip E. Auron, Matthew L. Hedberg, Lin Wang, Jeffery J. Raizer, John A. Kessler, Andrew T. Parsa, Wei-Qiang Gao, Sung-Hak Kim, Mutsuko Minata, Ichiro Nakano, Jennifer R. Grandis, Roger E. McLendon, Darell D. Bigner, Hui-Kuan Lin, Frank B. Furnari, Webster K. Cavenee, Bo Hu, Hai Yan and Shi-Yuan Cheng

Supplemental Figures 1-24 and Supplemental Tables 1-2

1 A B 20 400 DCBLD2 ST3GAL6 10 300 Copy number 0 200

99.9 100 100.1 100 Chr 3 (Mb) ST3GAL6 DCBLD2 0 1 4 7 10 13 16 19 22 25 28 Expression in GBM (relative fold)

C D

20 P<0.05 20 P<0.05 *

10 10 Tags/200,000 bp Tags/200,000 Tags/200,000 bp Tags/200,000 0 NormalGBM Normal GBM 0 Normal Grade II Grade III GBM ST3GAL6 DCBLD2

E F

Log2 (Tumor/normal) 0 Neural Proneural Classical Mesenchymal

Supplementary Figure 1. Genomic expression of DCBLD2 gene is up-regulated in clinical GBMs. (A) Digital karyotyping reveals a focal region on chromosome 3q12.1 displaying high copy numbers that indicate gene amplification. This amplified region includes two genes, ST3GAL6 and DCBLD2. (B) Quantitative real time reverse transcription PCR (Q-PCR) identifies increased expression of DCBLD2 but not ST3GAL6 in 14 out of 28 clinical GBM tumors. (C) Serial analysis of gene expression (SAGE) of DCBLD2 and ST3GAL6 in GBMs. Box, 25th-75th percentile with median. Whiskers, minimum and maximum values. (D) SAGE of DCBLD2 in subsets of gliomas. Normal vs. GBM, P<0.05. Box, 25th-75th percentile with median. Whiskers, minimum and maximum values. (E) Genomic expression of DCBLD2 among GBM subtypes (n=116). Number of GBM tumors, neural, 19, proneural, 37, classical, 22 and mesenchymal, 38. P<0.001 for mesenchymal vs. neural, mesenchymal vs. proneural, mesenchymal vs. classical. (F) Interphase FISH in a clinical GBM tumor TB2580 identifies tumor cells with amplification of DCBLD2 (green, FITC) compared to a chromosome 3 reference control (red, TRITC). Nuclei are stained blue (DAPI). Red arrows, DCBLD2 probe. Yellow arrows, Chromosome 3 reference probe. Overlap of DCBLD2 and chromosome 3 probes produces white-dish color indicated by yellow arrows. Data are derived from The Cancer Genome Anatomy Project (C and D) or The Cancer Genome Atlas (TCGA, E).

A B

SNB19U87LN229T98GD54LN444LN443LN319LN340 PCI-15BUM22ACal-33OSC-19 EGFR EGFR DCBLD2 DCBLD2 TRAF6 TRAF6 -actin -actin

C D

A549201T343TH23H1650H3255 A37516082TPF-11-74UACC903 EGFR EGFR DCBLD2 DCBLD2 TRAF6 TRAF6 -actin -actin

Supplementary Figure 2. Expression of EGFR, DCBLD2, and TRAF6 in cell lines derived from human glioma, lung cancer, HNC and melanoma. A. Expression of EGFR, DCBLD2 and TRAF6 in nine glioma cell lines. B. Expression of EGFR, DCBLD2 and TRAF6 in four HNC lines. C. Expression of EGFR, DCBLD2 and TRAF6 in six lung cancer cell lines. D. Expression of EGFR, DCBLD2 and TRAF6 in four melanoma cell lines. β-actin was used as a loading control. Data are representative of three independent experiments.

3 A B

shC shD2 #1 shD2 #2 1.2

shC shD2 #1 shD2 #2 shC shD2 #1 shD2 #2 shC shD2 #1 shD2 #2 0.8 DCBLD2 0.4 -actin U87 SNB19 LN444 0

Cell proliferation (Log2) U87 SNB19 LN444

U87 /Parental shC shD2 #1 shD2 #2 H&E

Supplementary Figure 3. Knockdown of endogenous DCBLD2 showed minimal effect on glioma cell proliferation in vitro and a modest impact on glioma tumor growth in the brain. A. Knockdown of endogenous DCBLD2 by two different shRNAs (shD2#1, shD2#2) in glioma U87, SNB19 and LN444 cells. IB analysis. β-actin was used as a loading control. B. Depletion of endogenous DCBLD2 did not affect in vitro cell proliferation of glioma cells. Cell proliferation of indicated glioma cells was determined by WST-1 assays using serum-starved glioma cells from panel A. Cells were seeded in 6 replicates. Error bars, ± SD. C. Inhibition of DCBLD2 by shRNA knockdown only displayed a modest impact on tumorigenesis of glioma U87 cells in the brain of mice. Representative images of H&E stained brain sections with indicated U87 gliomas from 5 mice per group of two independent experiments. Scale bars, 100 µm. Data are representative of two to three independent experiments.

A P/shC vIII/shC P/shC vIII/shC

vIII/shD2 #1 vIII/shD2 #2 vIII/shD2 #1 vIII/shD2 #2

SNB19 U87

B C Glima Stem Cells Glioma Stem Cells #83 #1123 #528 #18(JK) #42(JK) #92(JK) #83(JK) #83 #1123 #528 #18(JK) #42(JK) #92(JK) #83(JK) EGFR EGFR EGFRvIII EGFRvIII -actin -actin

Supplementary Figure 4. DCBLD2 is required for EGFR-driven glioma tumorigenesis.

(A) Effect of DCBLD2 knockdown by shD2#1, shD2#2 or shC on glioma cell colony formation. Cells were seeded on soft agar in triplicate. Scale bars, 1 mm. (B) & (C) Analyses of gene expression of WT EGFR or EGFRvIII in various patient- derived glioma stem cells (GSC) by RT-PCR using primers that specifically distinguish gene products of WT or EGFRvIII (1) (B) or by IB using an anti-EGFR antibody (C). WT EGFR were detected in GSC528, JK18, JK42 and JK 83 whereas EGFRvIII is expressed in GSC83, 1123 and JK92 cells. Data are representative of two independent experiments.

EGF - + - + - + p-Y-DCBLD2 IP DCBLD2 DCBLD2 p-Y-EGFR Whole Cell -actin Lysate 343T PCI-15B A375

Supplementary Figure 5. EGF stimulates p-Y-DCBLD2 in cell lines derived from lung cancer, HNC and melanoma. IP-IB assays. EGF stimulates p-Y-DCBLD2 in lung cancer 343T, HNC PCI-15B and melanoma A375 cells. Immunoprecipitated p-Y-DCBLD2 was detected with a pan anti- tyrosine antibody, 4G10. p-Y-EGFR was detected with an anti-p-EGFRY1045 antibody. DCBLD2 and β-actin were used as loading controls. Data are representative of three independent experiments.

Erlotinib - - + EGF - + + EGFR IP DCBLD2 DCBLD2 DCBLD2 IP EGFR EGFR

Supplementary Figure 6. Reciprocal IP-IB analyses failed to detect DCBLD2 association to EGFR in glioma cells. IP-IB assays. Immunoprecipitated DCBLD2 (upper panels) or EGFR (lower panels) was subjected to IB analyses of EGFR and DCBLD2 from U87/EGFR WT cells stimulated with EGF (50 ng/ml, 10 min) with or without pretreatment of EGFR inhibitor, Erlotinib (10 µM, 1h). Data are representative of three independent experiments.

Supplementary Figure 7. Mutant forms of DCBLD2F621, DCBLD2F750, and DCBLD2F621/F750 have no effect on EGFRvIII-stimulated Erk1/2 activity. IB analyses of Erk1/2 activity. Re-expression of Flag-DCBLD2 shRNA-resistant DCBLD2WT (WT) or indicated mutants or a vector control does not affect p-Erk1/2 in U87/EGFRvIII/shD2 cells. Erk1/2 and β-actin were used as loading controls. Data are representative of three independent experiments.

A B shControl+C shD2+C shD2+WT 120 ** shControl shD2 shD2+WT 80 shD2+F621 shD2+F750 shD2+F621 shD2+F750 shD2+F621/F750 shD2+F621/F750 40

U87/EGFRvIII/shD2

% soft agar colony 0

Supplementary Figure 8. Re-expression of Flag-DCBLD2 shRNA-resistant DCBLD2WT or DCBLD2F621 but not DCBLD2F750, DCBLD2F621/F750 mutant, or a vector control (C), rescues EGFRvIII-stimulated soft agar colony formation in vitro. A. Soft agar colony formation assay for U87/EGFRvIII/shD2 that separately re-express shRNA-resistant WT or mutant forms of DCBLD2. Representative images of three independent experiments. Scale bars, 1 mm. B. Quantifications of soft agar colonies. Error bars from 3 replicates, ± SD. Compared to shControl, **, P < 0.01. Data are representative of three independent experiments.

A B

Flag-DCBLD2- ControlWTF621 F750 F621/F750 shC shD2shD2 #1 shC #2 shD2shD2 #1 #2 p-Y750 Y750 IP p-DCBLD2 Flag DCBLD2 Flag -actin DCBLD2 Total cell U87/P U87/vIII -actin lysate C U87/vIII/shD2

Blocking Peptide - +

p-DCBLD2Y750

Supplementary Figure 9. Validation of the specificity of the anti-p-DCBLD2Y750 antibody. A. IB assays. EGFRvIII activates p-DCBLD2Y750 that is inhibited by shRNA knockdown of DCBLD2. The EGFRvIII-stimulated p-DCBLD2Y750 was detected with a rabbit anti-p- DCBLD2Y750 antibody generated against a specific phospho-peptide containing p-Y750 and surrounding amino acids. β-actin was used as a loading control. B. Re-expression of shRNA-resistant DCBLD2WT and mutant DCBLD2F621, but not DCBLD2F750 or DCBLD2F621/F750, rescues the EGFRvIII-stimulated p-DCBLD2Y750 in glioma U87/EGFRvIII/shD#2 cells. β-actin was used as a loading control. C. IHC assays of human a clinical GBM tumor tissue with the specific anti-p-DCBLD2Y750 antibody in the presence or absence of a specific blocking peptide synthesized with identical amino acid sequence of an immunoactive peptide for generating this specific anti-p-DCBLD2Y750 antibody including the Y750 residue. IHC was performed twice on the GBM sample with the blocking peptide with similar results. Scale bars, 50 µm. Data are representative of two to three independent experiments.

Supplementary Figure 10. EGFR phosphorylates DCBLD2 at Y750 in vitro. In vitro kinase phosphorylation assay. A Flag-tagged DCBLD2WT (WT) or a DCBLD2F750 mutant was separately expressed in HEK293T cells and immunoprecipitated with an anti-Flag antibody. The immunoprecipitated DCBLD2 proteins were pre-treated with a recombinant protein tyrosine phosphatase (PTP) and then incubated with or without a recombinant active EGFR. The reaction mixtures were examined by IB analysis using the specific anti-p-DCBLD2Y750 antibody. Flag-DCBLD2 was used as a loading control. Data are representative of two independent experiments.

SNB19343T PCI-15BA375 PDGF-A - + - + - + - + ControlHGFEGFPDGF-A p-Y750 p-Y-PDGF SNB19 DCBLD2 PDGFR HGF - + - + - + - + p-Y750 IP: p-Y-MET 343T DCBLD2 MET p-Y750 DCBLD2 PCI-15B DCBLD2 EGF - + - + - + - + p-Y-EGFR p-Y750 EGFR A375 DCBLD2 -actin

Supplementary Figure 11. p-DCBLD2Y750 is stimulated by EGF but not HGF or PDGF-A in various types of human cancer cells. Glioma SNB19, lung cancer 343T, HNC PCI-15B and melanoma A375 cells were serum- starved for 24 h and treated with or without 50 ng/ml PDGF-A, 100 ng/ml EGF, or 40 ng/ml HGF for 5 min. A specific anti-p-DCBLD2Y750 antibody was used to detect p-Y of endogenous DCBLD2 in these cells; anti-c-Met (p-Y1230/Y1234/Y1235), anti-p- PDGFRα (p-Y754) and anti-p-EGFR (p-Y1045) antibodies were used to examine p-Y of c-Met, PDGFRα and EGFR, respectively. DCBLD2, PDGFRα, EGFR, and β-actin were used as loading controls. Data are representative of three independent experiments.

10 A B

GBM39GBM14GBM6 EGFR shC shD2shC shD2 DCBLD2 DCBLD2 TRAF6 p-Akt-T308 -actin Akt -actin DCBLD2 IP GBM39 GBM6 p-DCBLD2-Y750 DCBLD2 TRAF6

C D shC shC shD2 shD2 1.5 150 * *

1.0 100

0.5 50

0 % soft agar colony 0 Cell proliferation (Log2) GBM39 GBM6 GBM39 GBM6

Supplementary Figure 12. Association of DCBLD2 with TRAF6 and effect of knockdown of DCBLD2 on cell proliferation in primary GBM cells with high levels of endogenous EGFRvIII expression. A. IP-IB analyses of association of DCBLD2 with TRAF6. Human primary short-term cultured glioma GBM6 and GBM39 cells express endogenous EGFRvIII at high levels. GBM14 has non-detectable EGFR WT or EGFRvIII proteins (2, 3). β-actin was used as a loading control. B. Knockdown of DCBLD2 by shRNA (shD2) inhibits EGFRvIII-stimulated Akt activity in primary GBM cells with endogenous EGFRvIII overexpression. Akt and β-actin were used as loading controls. C. Knockdown of DCBLD2 attenuates cell proliferation of GBM6 and GBM39 cells with high levels of endogenous EGFRvIII expression. Cells were seeded in 6 replicates. D. Knockdown of DCBLD2 inhibits soft-agar colony formation of GBM6 and GBM39 cells. Cells were seeded on soft agar in three replicates. In panels C and D, error bars, ± SD. Compared with shC-treated cells, *, P<0.05. Data are representative of three independent experiments.

Ni-NTA His-Ub + + + EGFRvIII - - + HA-TRAF6 - + +

HA IP: HA IB: TRAF6 (Ub)n

HA Whole EGFR Cell Lysate -actin

Supplementary Figure 13. EGFRvIII stimulates TRAF6 E3 ligase activity. IP-IB analyses. HA-TRAF6 and His-Ub were co-expressed with or without EGFRvIII in HEK293T cells. His-Ub was precipitated by Ni-nitrilotriacetic acid (NTA). β-actin was used as a loading control. Data are representative of three independent experiments.

EGFRvIII - - + + Flag-DCBLD2-WT + + + + HA-TRAF6 WT C70A WT C70A HA p-Y750 IP Flag Flag

HA Flag Whole Cell EGFR Lysate -actin

HEK293T

Supplementary Figure 14. Loss of TRAF6 E3 ligase activity has no effect on EGFRvIII-induced association of TRAF6 with DCBLD2. IP-IB analyses. Flag-DCBLD2 and HA-TRAF6 WT or a C70A mutant was co-expressed with or without EGFRvIII in HEK293T cells. HA-TRAF6, Flag-DCBLD2, and β-actin were used as loading controls. Data are representative of three independent experiments.

MDA-MB-231BT474MDA-MB-468SUM149SKBr3U87 SNB19T98GLN444 DCBLD2 TRAF6 Skp2 -actin Breast Cancer Glioma

Supplementary Figure 15. Expression of DCBLD2, Skp2, and TRAF6 in human breast cancer and glioma cell lines. IB analyses. Cell lysates of human cancer cell lines were examined for expression of DCBLD2, TRAF6 and Skp2 with corresponding antibodies. DCBLD2 was detected at low levels in four of five breast cancer cell lines examined. In contrast, DCBLD2 is expressed at high levels in all four glioma cell lines examined here and nine glioma cell lines (including these four) shown in Supplemental Figure 2A. β-actin was used as a loading control. Data are representative of two independent experiments.

A WT -/-

TRAF6TRAF6 DCBLD2

TRAF6 -actin MEF

B C

ControlTRAF6 siRNAControl siRNATRAF6 siRNA siRNA shControlshTRAF6shTRAF6 #1 shTRAF6 #2shTRAF6 #3 #4 DCBLD2 DCBLD2 TRAF6 TRAF6 -actin -actin SNB19 U87 U87/EGFRvIII

Supplementary Figure 16. Inhibition of TRAF6 expression decreases DCBLD2 expression in mouse embryonic fibroblasts (MEF) and glioma cells. IB analyses of expression of DCBLD2 and TRAF6.

A. Expression of endogenous DCBLD2 was at very low levels in TRAF6 null (TRAF6-/-) MEF but at high levels in TRAF6 wild type (TRAF6WT) MEF. B. Expression of endogenous DCBLD2 was significantly decreased when endogenous TRAF6 was knocked down by siRNAs in glioma SNB19 and U87 cells. C. Expression of endogenous DCBLD2 was markedly decreased when endogenous TRAF6 was knocked down by two separate shRNAs, #1 and #3, but not shRNAs #2 and #4, or a control shRNA in glioma U87/EGFRvIII cells. β-actin was used as a loading control. Data are representative of two independent experiments.

A B WT WT GFP D750 GFP D750 DCBLD2 DCBLD2 -actin -actin shC shD2 shC shD2 U87/vIII GSC#83

Supplementary Figure 17. Re-expression of DCBLD2 WT or DY750 mutant in glioma U87 cells or patient-derived glioma stem cells (GSC#83) that endogenous DCBLD2 was knocked down by shRNA for DCBLD2 (shD2). IB analyses. Cell lysates of various U87 or patient-derived glioma stem cells (GSC#83) (4) were examined for expression of DCBLD2WT or DCBLD2D750 mutant. shC, cells expressed a scrambled control shRNA. shD2, cells expressed a shRNA for DCBLD2. GFP, cells expressed GFP, WT, cells express DCBLD2WT and D750, cells expressed DCBLD2D750 mutant. β-actin was used as a loading control. Data are representative of two independent experiments.

Supplementary Figure 18. Examples of a clinical GBM sample showing co-expression of EGFR, p-EGFRY1172, p-DCBLD2Y750, TRAF6, and p-AktT308 and a GBM sample with negative staining for these proteins. Representative images of two GBM specimens that were IHC stained positive (J213) or negative (J45) by indicated antibodies. Scale bars, 50 µm. IHC analyses on these two GBM samples using these five antibodies were performed twice. Data are representative of two independent experiments.

Supplementary Figure 19. Examples of two clinical HNC samples showing co- expression of EGFR, p-EGFRY1172, p-DCBLD2Y750, TRAF6, and p-AktT308 and two HNC samples with negative staining for these proteins. Representative images of two HNC specimens that were IHC stained positive (P32-1AFS and HN11-6062) and two HNC tissues that showed negative staining (P7 and HN11-6026-3J) by indicated antibodies. Scale bars, 100 µm. IHC analyses on these HNC samples in TMAs using these five antibodies were performed twice. Data are representative of two independent experiments.

A GBM 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 EGFR p-DCBLD2Y750 TRAF6 p-AktT308 -actin

B HNC

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 T N T N T N T N T N T N T N T N T N T N T N T N T N T N T N EGFR p-DCBLD2Y750 TRAF6 p-AktT308 -actin

Supplementary Figure 20. Expression of EGFR, p-DCBLD2Y750, TRAF6 and p-Akt in separate cohorts of 19 GBM specimens, 15 HNC samples and their matched normal tissues. IB analyses of expression of EGFR, p-DCBLD2Y750, TRAF6 and p-AktT308 in 19 snap-frozen clinical GBM specimens (A), 15 snap-frozen HNC samples (T) and their matched normal tissues (N) (B). β-actin was used as a loading control. Data are representative of 2 independent experiments.

120 120 Low p-EGFRY1172 (n=74) Low p-DCBLD2Y750 (n=66) High p-EGFRY1172 (n=58) High p-DCBLD2Y750 (n=66) 90 90

P<0.05 P<0.05 60 60

30 30 Percent survival Percent survival

0 0 0 50 100 150 0 50 100 150 Months Months

120 120 Low TRAF6 (n=38) Low p-EGFRY1172/TRAF6 (n=88) High TRAF6 (n=96) High p-EGFRY1172/TRAF6 (n=44) 90 90

P<0.05 60 60

30 30 Percent survival Percent survival

0 0 0 50 100 150 0 50 100 150 Months Months

Supplementary Figure 21. High expression of either p-EGFRY1172 or p-DCBLD2Y750, or co-expression of p-EGFRY1172 with TRAF6, correlates with worse survival of patients with gliomas. Kaplan-Meier analyses of patients with high p-EGFRY1172, p-DCBLD2Y750-expressing, or co- expression of p-EGFRY1172-TRAF6 tumors (blue line) versus low p-EGFRY1172, p- DCBLD2Y750-expressing or low co-expression of p-EGFRY1172-TRAF6 tumors (red line) in IHC-stained WHO grades II-IV glioma specimens showed in Figure 8A and Supplemental Table 1. Expression of TRAF6 does not correlate with clinical outcomes. P values were calculated by using log-rank test. Black bars, censored data.

120 120 Y750 Low p-EGFRY1172 (n=166) Low p-DCBLD2 (n=209) Y750 Y1172 High p-DCBLD2 (n=22) 90 High p-EGFR (n=65) 90

60 60 P<0.01 30 30 Percent survival Percent survival

0 0 0 50 100 150 200 250 0 50 100 150 200 250 Months Months

120 Low p-EGFRY1172/TRAF6 (n=157) High p-EGFRY1172/TRAF6 (n=74) 90

60 P<0.05

30 Percent survival

0 0 50 100 150 200 250 Months

Supplementary Figure 22. High expression of p-EGFRY1172 or co-expression of EGFR with TRAF6 correlates with worse survival of patients with HNCs. Kaplan-Meier analyses of patients with high expression of p-EGFRY1172, p-DCBLD2Y750 or co-expression of p-EGFRY1172-TRAF6 tumors (blue line) versus low p-EGFRY1172 p- DCBLD2Y750 or p-EGFRY1172-TRAF6-expressing tumors (red line) in IHC-stained TMA of HNCs showed in Figure 8C and Supplemental Table 2. Expression of p-DCBLD2Y750 does not correlate with clinical outcomes. P values were calculated by using log-rank test. Black bars, censored data.

A B

EGFRvIII p-DCBLD2Y750

Low EGFRvIII/p-DCBLD2Y750 (n=132) High EGFRvIII/p-DCBLD2Y750 (n=16)

P<0.05 TRAF6 p-AktT308 Percent survival

Months

Supplementary Figure 23. EGFRvIII, p-DCBLD2Y750, TRAF6, and p-AktT308 are co- expressed in clinical gliomas and co-expression of EGFRvIII with p-DCBLD2Y750 correlates with poor survival of patients with gliomas A. A total of 148 clinical primary glioma specimens including WHO grade I to IV tumors were analyzed by IHC for p-DCBLD2Y750, TRAF6, p-AktT308 and EGFRvIII. Representative images of serial sections of a grade IV GBM tissue using anti- EGFRvIII (clone 8.3), anti-p-DCBLD2Y750, anti-TRAF6, and anti-p-AktT308 antibodies are shown. Inserts, isotype-matched IgG controls of the same area in adjacent sections. Arrows, positive staining. Scale bars, 50 µm. IHC analyses on these glioma tumor samples using these four antibodies were performed twice. Data are representative of two independent experiments. B. Kaplan-Meier analyses of patients with high EGFRvIII/p-DCBLD2Y750-expressing tumors (blue line) versus low EGFRvIII/p-DCBLD2Y750-expressing tumors (red line) in IHC staining assays (A) of WHO grades I-IV gliomas. P values were calculated by using log-rank test. Black bars, censored data.

Supplementary Figure 24. The levels of expression of p-DCBLD2Y750 and p-EGFRY1172 -are increased in WHO tumor grade II to IV clinical gliomas compared with normal brain tissues. IHC staining of 31 WHO grade II, 23 grade III and 78 grade IV (GBM) glioma specimens using anti-EGFRY1172 or anti-p-DCBLD2Y750 antibodies with validated specificities. The data were compared with four IHC-stained normal brains that had no detectable pathological lesions. *, P < 0.05, **, P < 0.01 and ***, P < 0.001.

Supplementary Table 1. Spearman’s rank correlation analysis of expression level of p-EGFRY1172, p-DCBLD2Y750, TRAF6 and p-AktT308 in human clinical glioma specimens by IHC staining.

r p-EGFRY1172 p-DCBLD2Y750 TRAF6 p-AKTT308 p-EGFRY1172 1.0c 0.5c 0.4c 0.2a p-DCBLD2Y750 0.5c 1.0c 0.2b 0.3c TRAF6 0.4c 0.2b 1.0c 0.5c p-AKTT308 0.2a 0.3c 0.5c 1.0c

Note: a. P<0.05; b. P<0.01 and c. P<0.001

Supplementary Table 2. Spearman’s rank correlation analysis of expression level of p-EGFRY1172, p-DCBLD2Y750, TRAF6 and p-AktT308 in human clinical HNC specimens by IHC staining.

r p-EGFRY1172 p-DCBLD2Y750 TRAF6 p-AktT308 p-EGFRY1172 1.0c 0.2a 0.4c 0.1 p-DCBLD2Y750 0.2a 1.0c 0.2b 0.3c TRAF6 0.4c 0.2b 1.0c 0.1 p-AKTT308 0.1 0.3b 0.1 1.0c

Note: a. P<0.05; b. P<0.01 and c. P<0.001

References:

1. Inda, M.M., et al. Tumor heterogeneity is an active process maintained by a mutant EGFR-induced cytokine circuit in glioblastoma. Genes Dev 2010; 24(16):1731-1745. 2. Giannini, C., et al. Patient tumor EGFR and PDGFRA gene amplifications retained in an invasive intracranial xenograft model of glioblastoma multiforme. Neuro-oncol 2005; 7(2):164-176. 3. Yiin, J.J., et al. ZD6474, a multitargeted inhibitor for receptor tyrosine kinases, suppresses growth of gliomas expressing an epidermal growth factor receptor mutant, EGFRvIII, in the brain. Mol Cancer Ther 2010; 9(4):929-941. 4. Mao, P., et al. Mesenchymal glioma stem cells are maintained by activated glycolytic metabolism involving aldehyde dehydrogenase 1A3. Proc Natl Acad Sci U S A 2013; 110(21):8644-8649.

25 Full unedited gels for Figure 1

A D

anti-DCBLD2 anti-DCBLD2

anti-p-EGFR anti-p-EGFR

anti-EGFR anti-EGFR

anti-β-actin anti-β-actin

Full unedited gels for Figure 2

A G

anti-EGFR

anti-DCBLD2

anti-DCBLD2 anti-β-actin

anti-β-actin

Rectangles indicate cropped areas used in indicated Figures Full unedited gels for Figure 3

A B

anti-p-DCBLD2 anti-p-DCBLD2 anti-p-DCBLD2

anti-DCBLD2 anti-DCBLD2 anti-DCBLD2

anti-p-EGFR anti-p-EGFR anti-p-EGFR

anti-EGFR anti-EGFR anti-EGFR

anti-β-actin anti-β-actin anti-β-actin

D E

anti-p-Y-Flag-DCBLD2 anti-DCBLD2

anti-Flag-DCBLD2 anti-p-AktT308

S473 anti-EGFR anti-p-Akt

anti-β-actin anti-Akt

Rectangles indicate cropped areas used in indicated Figures Full unedited gels for Figure 4 A B C

anti-HA anti-HA anti-TRAF6 anti-p-Y750 anti-p-Y750

anti-DCBLD2 anti-Flag

anti-Flag anti-Flag anti-DCBLD2

anti-HA anti-HA anti-TRAF6

anti-HA anti-Flag

anti-Flag anti-EGFR anti-EGFR

anti-β-actin anti-β-actin D E

anti-HA anti-TRAF6

anti-p-Y750 anti-Flag

anti-Flag anti-TRAF6

anti-HA anti-DCBLD2

anti-EGFR anti-β-actin

Rectangles indicate cropped areas used in indicated Figures anti-β-actin Full unedited gels for Figure 5 A B C

anti-HA anti-HA anti-HA

Run in different anti-HA gels with the anti-Flag same samples anti-Flag

anti-HA anti-Flag anti-HA

anti-EGFR anti-EGFR anti-EGFR

anti-β-actin anti-β-actin anti-β-actin

anti-Ub

anti-TRAF6

Rectangles indicate cropped areas used in indicated Figures Full unedited gels for Figure 6

A C

anti-p-AktT308 anti-TRAF6

anti-Akt

anti-β-actin

D E

anti-HA anti-Akt

anti-p-AktT308 anti-Akt

anti-β-actin

anti-p-AktT308

anti-β-actin Full unedited gels for Figure 7 A C

anti-p-Y-EGFR anti-p-Y-EGFR

anti-EGFR anti-EGFR

anti-Flag (DCBLD2)

anti-Flag (DCBLD2) anti-p-AktT308

anti-p-AktT308 anti-Akt

anti-Akt anti-TRAF6 IP: anti-Flag D2

anti-TRAF6 IP: anti-Flag D2 anti-Flag (DCBLD2)

anti-Flag (DCBLD2)

IP: IP: anti-TRAF6 anti-Flag (DCBLD2) anti-TRAF6 anti-Flag (DCBLD2)

anti-TRAF6 anti-TRAF6 Full unedited gels for Supplemental Figure 2

A B

anti-EGFR anti-EGFR

anti-DCBLD2 anti-DCBLD2

anti-TRAF6 anti-TRAF6

anti-β-actin anti-β-actin

C D

anti-EGFR anti-EGFR

anti-DCBLD2 anti-DCBLD2

anti-TRAF6 anti-TRAF6

anti-β-actin anti-β-actin

Rectangles indicate cropped areas used in indicated Figures Full unedited gels for Supplemental Figure 3

A U87 SNB19 LN444

anti-DCBLD2 anti-DCBLD2 anti-DCBLD2

anti-β-actin anti-β-actin anti-β-actin Full unedited gels for the indicated Figures

Supplemental Figure 4C Supplemental Figure 5 Supplemental Figure 6

anti-EGFR anti-Y-DCBLD2

anti-DCBLD2 anti-DCBLD2 anti-EGFR

anti-Y-EGFR anti-DCBLD2

anti-β-actin anti-EGFR anti-β-actin

Supplemental Figure 9

anti-Y-DCBLD2 Y750

anti-DCBLD2 Supplemental Figure 7

anti-β-actin

anti-p-Erk1/2 B

anti-Erk1/2

anti-p-Y750

anti-β-actin anti-Flag

anti-DCBLD2

anti-β-actin Rectangles indicate cropped areas used in indicated Figures Full unedited gels for the indicated Figures

Supplemental Figure 10 Supplemental Figure 11

anti-p-Y750 anti-p-Y-PDGFRα

anti-DCBLD2 anti-p-Y750 anti-PDGFRα

anti-p-Y750

anti-Flag anti-p-Y-MET anti-DCBLD2

anti-p-Y750 anti-MET

anti-p-Y-EGFR anti-DCBLD2

anti-EGFR anti-p-Y750

anti-β-actin

anti-DCBLD2

Rectangles indicate cropped areas used in indicated Figures Full unedited gels for the indicated Figures

Supplemental Figure 12 Supplemental Figure 13

A B

anti-EGFR anti-DCBLD2 anti-HA

anti-DCBLD2 anti-p-AktT308 anti-HA

anti-TRAF6 anti-p-Akt anti-EGFR

anti-β-actin anti-β-actin anti-β-actin Supplemental Figure 14

anti-DCBLD2 Supplemental Figure 15 anti-HA

anti-Y-DCBLD2Y750 anti-p-Y750 anti-DCBLD2

anti-TRAF6 anti-Flag anti-TRAF6

anti-HA anti-Spk2 anti-Flag

anti-β-actin

anti-EGFR

anti-β-actin

Rectangles indicate cropped areas used in indicated Figures Full unedited gels for the indicated Figures

Supplemental Figure 16

A B C

anti-DCBLD2 anti-DCBLD2 anti-DCBLD2

anti-TRAF6 anti-TRAF6 anti-TRAF6

anti-β-actin anti-β-actin anti-β-actin

Supplemental Figure 17

A B

anti-DCBLD2 anti-DCBLD2

anti-β-actin anti-β-actin

Rectangles indicate cropped areas used in indicated Figures Full unedited gels for Supplemental Figure 20

anti-EGFR anti-EGFR

anti-p-DCBLD2Y750 anti-p-DCBLD2Y750

anti-TRAF6 anti-TRAF6

T308 anti-p-Akt anti-p-AktT308

anti-β-actin anti-β-actin B

anti-EGFR anti-p-DCBLD2Y750 anti-TRAF6

anti-p-AktT308 anti-β-actin

Rectangles indicate cropped areas used in the indicated Figures.