Supplementary Information
Acquired MET Expression
Confers Resistance to EGFR Inhibition
In a Mouse Model of Glioblastoma Multiforme
Hyun Jung Jun, Jaime Acquaviva, Dorcas Chi, Julie Lessard, Haihao Zhu, Steve Woolfenden, Roderick T.
Bronson, Rolf Pfannl, Forest White, David E. Housman, Lakshmanan Iyer, Charles A. Whittaker, Abraham
Boskovitz, Ami Raval, Alain Charest
Supplementary Table 1. Information on the antibodies used for immunohistochemical characterization of
mouse GBM tumors.
Supplementary Table 2. Information on antibodies used in immunoblot analyses.
Supplementary Table 3. GSEA of mouse GBM tumors.
Size; gene set size, ES; enrichment score, NES; normalized enrichment score, NOM p-val; normalized p value,
FDR q-val; false discovery rate, FWER p-val; family-wise error rate.
Supplemental Table 4. Gene list information for Taqman assays.
References
Charest A, Wilker EW, McLaughlin ME, Lane K, Gowda R, Coven S et al (2006). ROS fusion tyrosine kinase
activates a SH2 domain-containing phosphatase-2/phosphatidylinositol 3-kinase/mammalian target of
rapamycin signaling axis to form glioblastoma in mice. Cancer Res 66: 7473-81.
Coleman JE, Huentelman MJ, Kasparov S, Metcalfe BL, Paton JF, Katovich MJ et al (2003). Efficient largescale production and concentration of HIV-1-based lentiviral vectors for use in vivo. Physiol Genomics 12: 2218.
Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S et al (2004). Bioconductor: open software
development for computational biology and bioinformatics. Genome Biol 5: R80.
Lesche R, Groszer M, Gao J, Wang Y, Messing A, Sun H et al (2002). Cre/loxP-mediated inactivation of the
murine Pten tumor suppressor gene. Genesis 32: 148-9.
Saeed AI, Bhagabati NK, Braisted JC, Liang W, Sharov V, Howe EA et al (2006). TM4 microarray software
suite. Methods Enzymol 411: 134-93.
Serrano M, Lee H, Chin L, Cordon-Cardo C, Beach D, DePinho RA (1996). Role of the INK4a locus in tumor
suppression and cell mortality. Cell 85: 27-37.
Shin KJ, Wall EA, Zavzavadjian JR, Santat LA, Liu J, Hwang JI et al (2006). A single lentiviral vector platform
for microRNA-based conditional RNA interference and coordinated transgene expression. Proc Natl Acad Sci
U S A 103: 13759-64.
Silver DP, Livingston DM (2001). Self-excising retroviral vectors encoding the Cre recombinase overcome Cremediated cellular toxicity. Mol Cell 8: 233-43.
Smyth GK (2004). Linear models and empirical bayes methods for assessing differential expression in
microarray experiments. Stat Appl Genet Mol Biol 3: Article3.
Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA et al (2005). Gene set enrichment
analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U
S A 102: 15545-50.
Tusher VG, Tibshirani R, Chu G (2001). Significance analysis of microarrays applied to the ionizing radiation
response. Proc Natl Acad Sci U S A 98: 5116-21.
Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD et al (2010). Integrated genomic analysis
identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR,
and NF1. Cancer Cell 17: 98-110.
Woolfenden S, Zhu H, Charest A (2009). A Cre/LoxP conditional luciferase reporter transgenic mouse for
bioluminescence monitoring of tumorigenesis. Genesis 47: 659-66.
Wu Z, Irizarry RA (2004). Preprocessing of oligonucleotide array data. Nat Biotechnol 22: 656-8; author reply
658.
Zhu H, Acquaviva J, Ramachandran P, Boskovitz A, Woolfenden S, Pfannl R et al (2009). Oncogenic EGFR
signaling cooperates with loss of tumor suppressor gene functions in gliomagenesis. Proc Natl Acad Sci U S A
106: 2712-6.
Supplementary Materials and Methods
EGFR Conditional Transgenic Mice
All mouse procedures were performed in accordance with Tufts University’s recommendations for the care and
use of animals and were maintained and handled under protocols approved by the Institutional Animal Care
and Use Committee. Cre/Lox-mediated conditional expression of the human wild type EGF receptor was
achieved by targeted knock in of CAG-floxed stop cassette EGFR cDNA mini gene (CAG-LSL-EGFRWT) into
the 3’ region of the mouse collagen 1α1 gene locus as described in details elsewhere (Zhu et al., 2009). The
Cre/Lox-mediated conditional expression of the firefly luciferase transgene Tg(CAG-luc)C6Char has been
described and characterized previously (Woolfenden et al., 2009). These mice were mated to p16Ink4a/p19Arf
null (Cdkn2atm1Rdp/tm1Rdp (Serrano et al., 1996)) and conditional PTENlox/lox knock out strains (Ptentm1Hwu/tm1Hwu
(Lesche et al., 2002)). The combinations of strains indicated in the text were produced by standard
crossbreedings. CAG-LSL-EGFRWT transgenic mice that were not exposed to Cre recombinase did not exhibit
phenotypic features that are consistent with spontaneous tumor formation (data not shown). Genotyping
protocols for Col1a1tm1(CAG-EGFR)Char, Cdkn2atm1Rdp/tm1Rdp, Tg(CAG-luc)C6Char and conditional Ptentm1Hwu/tm1Hwu
strains were carried as described elsewhere (Lesche et al., 2002; Serrano et al., 1996; Woolfenden et al.,
2009; Zhu et al., 2009).
Nomenclature of Strains
Throughout the manuscript, constitutive homozygous knock out Cdkn2atm1Rdp/tm1Rdp mice are referred to
Ink2/3-/- and conditional Ptentm1Hwu/tm1Hwu are referred to as PTENlox/lox when non-floxed or PTEN-/- when floxed
out. Animals carrying the CAG-LSL-EGFRWT transgene knock-in allele are indicated as CAG-LSL-EGFRWT.
Floxed alleles are referred to as EGFRWT.
Stereotactic Injections
Stereotactic injections of adult animals (3 months of age and above) have been described in details previously
(Charest et al., 2006; Zhu et al., 2009). Briefly, mice of the indicated genotype were anesthetized with an
intraperitoneal injection of Ketamine/Xylazine and mounted on a stereotaxic frame with non-puncturing ear
bars. The incision site was shaved and sterilized and a single incision is made from the anterior pole of the
skull to the posterior ridge. A burr hole was drilled at defined location and either a 1 μL Hamilton syringe or a
pulled glass pipet mounted onto a Nanoject II injector (Drummond Scientific Company) was used to inject the
Cre virus. Following retraction of the syringe or pipette, the burr hole was filled with sterile bone wax, the skin is
drawn up and sutured and the animal is placed in a cage with a padded bottom atop a surgical heat pad until
ambulatory.
GBM Primary Cultures Preparation
Primary cultures of tumors were established as follows: tumors were excised and minced in 0.25% trypsin
(wt/vol) 1mM EDTA and allowed to disaggregate for 15 minutes at 37oC. The resulting cell suspension was
then strained through a 70μm cell strainer (Falcon). The single suspension of cells was washed in PBS twice
and plated on 0.2% gelatin coated tissue culture plates. Cells were fed every 24 hours with fresh media that
consisted of DMEM supplemented with 10% heat inactivated fetal bovine serum and antibiotics.
Plasmids Construction, Virus Production and Titer Determination
PTEN shRNA screening. 4 shRNA sequences were chosen from TRC library website and cloned into pLKO.1
pgk-puro according to TRC protocols available on their website. After sequencing validation, the lentiviruses
were transitenly produced as described below and use to infect TGF-EGFRWT;InkΔ2/3-/- tumor cell cultures.
Following selection in puromycin (3.5 g/mL) pools of selected cells were assayed for PTEN expression by
immunoblot analysis. Mouse c-met gene promoter cloning and luciferase constructs. A 3.5 kb fragment of the
mouse c-met gene promoter region that includes the transcriptional initiation site was amplified by PCR using
the RP23-179P9 BAC (BacPac Resources) as a target sequence and the following primers 5’AAAAACTCGAGGCTGGTAGACAAGAAAACTTCCCAG-3’ and 5’-TTTCACCTCCA CTGAGTCCCA-3’ and
cloned into pZeroBlunt Topo (Invitrogen) and digested with XhoI and BamHI, gel purified and ligated into
pGL4.10[luc2] (Promega) vector digested with XhoI and BlgII. Cre Lentiviruses. The pTyf TGF-IRES-iCre and
pTyf eGFP-IRES-iCre lentivirus plasmids were constructed using standard molecular biology techniques and
their assembly detailed in Supplemental Experimental Procedures. High titer lentiviruses were produced by
transient transfections in HEK293T cells as follows: HEK293T cells were seeded in 12 X 10-cm2 culture plates
at a density of 4x107 cells per plate and maintained in DMEM supplemented with 10% FBS and antibiotics. For
each 10-cm2 culture plates, lentivirus was produced by co-transfection of 10ug lentiviral transducing vector,
7.5ug of ΔR8.9 packaging vector, and 5ug of VSV-G envelope vector using TransIT-LT1 transfection reagent
(Mirus) according to manufacturer’s recommendations. After 24 and 48hr post-transfection the virus-containing
media collected and pooled conditioned media was concentrated using Amicon filters (Millipore) according to
the manufacturer’s protocol to a final volume of 35 mL. The viral stock was further concentrated by ultracentrifugation at 27,000 rpm for 1.5 hrs and the supernatant was aspirated with caution. The concentrated
virus was resuspended in 0.5ml PBS, aliquoted and stored at -80ºC. To determine a functional titer, a dilution
series of pTyf TGF-IRES-iCre and pTyf eGFP-IRES-iCre high titer viral preparations (described above) were
used to infect a Cre/LoxP conditional 3T3-LacZ reporter cell line (Silver and Livingston, 2001) using 8 g/mL of
polybrene. 48 hours post infection, cells were washed with PBS, fixed in 0.5% glutaraldehyde in PBS for 5 min,
washed twice with PBS, and then incubated with 2 mM MgCl2, 5 mM K3Fe(CN)6, 5 mM K4Fe(CN)6, and 1
mg/ml X-gal in PBS overnight at 37oC in the dark. For each dilutions, the number of X-gal positive cells were
counted and the results converted to transducing units (TU) per volume (TU/mL).
Histology and Immunohistochemistry
Tumor-bearing animals were transcardially perfused with cold PBS and their brains were excised, rinsed in
PBS, and serial coronal sections cut using a brain mold. Half of the sections were used to isolate primary
cultures of tumor cells as described below and the other half were post-fixed in 4% paraformaldehyde,
embedded in paraffin, sectioned (5-10 μM) and stained with hematoxylin and eosin (H&E) (Sigma). For IHC,
cut sections were deparaffinized and rehydrated through xylenes and graded alcohol series and rinsed for 5
minutes under tap water. Antigen target retrieval solution (Dako, S1699) was used to unmask the antigen
(microwaved for 10 minutes at low power then cooled down for 30 minutes) followed by 3 washes with PBS for
5 minutes each. Quenching of endogenous peroxidase activity was performed by incubating the sections for 30
minutes in 0.3% H2O2 in methanol followed by PBS washes. Slides were pre-incubated in blocking solution
(5% (v/v) goat serum (Sigma) in PBS/0.3% (v/v) Triton-X100) for 1 hour at room temperature; followed by
mouse on mouse blocking reagent (Vector lab, Inc, MKB-2213) incubation for 1 hour. Primary antibodies
(details listed in Supplementary Table 1) were incubated for 24 hour. Secondary antibodies used were
biotinylated anti-rabbit or anti-mouse (Vector Labs, 1:500) for IHC. All antibodies were diluted in blocking
solution. All immunobinding of primary antibodies were detected by biotin-conjugated secondary antibodies
and Vectastain ABC kit (Vector lab, Inc) using DAB (Vector lab, Inc) as a substrate for peroxidase and
counterstained with hematoxylin. The Ki-67 proliferation index was calculated as the mean number of Ki-67
positive cells per high power field with the control representing 100%. The total number of tumor cells per high
power field is constant. TUNEL (Millipore apoptag plus peroxidase in situ apoptosis detection kit) positivity
index was calculated as mean number of TUNEL positive cells/total number of tumor cells per high power field
x 100.
Immunoblots
Western blots were performed as follows: cell lysates were prepared using RIPA buffer supplemented with 5
mM Na3VO4 (freshly made) and CompleteTM protease inhibitor cocktail (Roche). Equiamount of total cell
lysates were separated by SDS-PAGE and electrotransfered to PVDF membrane (Immobilon P, Millipore).
Blots were blocked in Tris-buffered saline 0.1% (v/v) Tween-20 (TBS-T), 1% (wt/v) BSA and 5% (wt/v) non fat
dry milk (Bio-Rad) for 1 hour on a shaker. Primary antibodies were added to blocking solution at 1:1000 dilution
and incubated overnight at 4oC on a shaker. Blots were washed several times with TBS-T BSA and secondary
antibodies were added at 1:10000 dilution into TBS-T BSA and incubated for 1 hour at room temperature on a
shaker. After several washed, enhanced chemiluminescence (ECL) reactions were performed as described by
the manufacturer (Western Lightning Kit, Perkin Elmer). The primary antibodies used in these studies are listed
in Supplementary Table 2.
Plasmid Construction and Virus Production
pTyf-GFP-IRES-iCre and pTyf-TGFα-IRES-iCre DNA constructs: The plasmid for pTyf-eGFP-IRES-iCre was
generated as follows: The cDNA for iCre was amplified by PCR from a iCre containing plasmid (a kind gift of
Dr.
Jason
Coleman,
MIT)
with
an
Mlu1
site-containing
forward
primer
(AAAAAAACGCGTCCACCATGGTGCCCAAGAAGAGGA)
and
a
reverse
primer
(TTTTTTATCGATTCAGTCCCCATCCTCGAGCAGC) using Pfu polymerase. The amplified PCR product was
digested with Mlu1, treated with CIP and cloned into Mlu1 EcoRV digested pTyf-IRES-eGFP transducing
vector backbone (Coleman et al., 2003) to generate pTyf-IRES-iCre. The cDNA for eGFP was amplified by
PCR from the pTyf-IRES-eGFP plasmid with a forward primer containing an Nhe1 site
(AAAAAAGCTAGCCCACCATGGTGAGC AAGGGCGAGGAGC) and a reverse primer containing a Cla1 site
(TTTTTTATCGATTTACTTGTACAGCTCGTCCATG) using Pfu polymerase. The amplified eGFP fragment was
cloned into the pCR-BluntII-TOPO shuttle vector (invitrogen) according to the manufacturer’s protocol. eGFP
was excised from the TOPO shuttle vector by digestion with Nhe1 and ligated into Nhe1-digested pTyf-IRESiCre to create pTyf-eGFP-IRES-iCre. In order to generate pTyf-TGFα-IRES-iCre the coding region for the
human TGFα cDNA was PCR amplified from the pcDNA3-TGFα vector (a kind gift from Dr. Robert Langley,
The University of Texas, M.D. Anderson Cancer Center, Houston) using Nhe1 containing primers forward
(AAAAAAAGCTAGCCGCGCAGGTAGGGCAGGAGGCT)
and
reverse
(TTTTTTGC
TAGCAAACTCCTCCTCTGG GCTCTTCA). The amplified TGFα fragment was digested with Nhe1 and ligated
into Nhe1-digested pTyf-IRES-iCre to generate pTyf-TGFα-IRES-iCre. All constructs were DNA sequenced for
integrity.
pSlik-PTEN-Blast: The previously described pSlik-Venus vector platform (Shin et al., 2006) for tetracyclineinducible transgene expression was modified to generate pSlik-PTEN-Blast for conditional expression of PTEN
and constitutive expression of a blasticidine resistance gene. Briefly, the pSlik-Venus vector was digested with
Kpn1 and HindIII and the the resulting 742 bp fragment was ligated into Kpn1/HindIII digested pBluescriptII to
create pBKSII-A. The cDNA for a blasticidine resistance gene was PCR amplified from a mammalian
expression system plasmid containing the Blast resistance gene with a forward primer containing tandem
HindIII and Age1 sites and a reverse primer containing a HindIII site. The amplified blasticidine resistance gene
was digested with HindIII and cloned into the HindIII site of pBKSII-A to create pBKSII-B. Next, pSlik-Venus
was digested with Age1 and the resulting 1.4kb fragment was ligated into Age1/Ava1 digested pBKSII-B to
generate pBKSII-C. Finally, pBSKII-C was digested with Asc1 and Bsu36I and the resulting 2.6kb fragment
containing the sequence for blasticidine resistance was ligated into Age1/Bsu36I digested pSlik-Venus to
create pSlik-Blast. The insertion of transgenes into the pSlik platform is based on Gateway recombination
cloning technology (Invitrogen). The cDNA for human PTEN was amplified from a PTEN containing vector with
primers PTEN-Fow-EcoRI (AAAAAGAATTCC ACCATGACAGCCATCATCAAAGAG) and PTEN-Rev-XhoI
(CCCCCCTCGAGTCAGA CTTTTGTAATTTGTG). The amplified PTEN cDNA was digested with EcoRI and
XhoI and ligated into EcoRI-XhoI digested pEN-Tmcs vector (Shin et al., 2006). The resulting pEN-PTEN
plasmid was used to shuttle PTEN into pSlik-Blast using Gateway LR clonase enzyme mix (Invitrogen)
according to the manufacturer’s protocols.
Crystal Violet Blue Staining Assay
For CV staining assays, cells were plated at 40,000 cells per well in a 12 well dishes and incubated overnight
in 0.1% FBS in media before the indicated treatments were performed. Post treatment, cells were fixed in 10%
formalin for 5 minutes and washed with PBS. Fixed cells were stained with 0.05% (wt/vol) crystal violet in
distilled water for 30 minutes and then washed with distilled water and drained. To read the amount of CV
stain, which is an indirect measure of cell number, CV was solubilized by adding 1 mL of methanol to each well
and incubated for 30 minutes. Solubilized CV was read spectrophotometrically at 540 wavelength. The results
are reported relative to untreated.
Survival Assays and Inhibitor Treatments
Cell viability was measured by a trypan blue exclusion assay. Cells were plated in triplicate at a density of
2x105 cells per well in a 12 well tissue culture plate and incubated for 8 hour at 37ºC in 5% CO2. Cells were
washed with PBS and serum starved for 16 hour in DMEM containing 0.1% FBS and antibiotics. Prior to
adding drug (T0), the total number of viable cells was determined by releasing cells from the well by
trypsinization and mixing equivalent volumes of cell suspension and 0.4% trypan blue. A hemacytometer was
used to count the total number of unstained, viable cells. For drug treatment, stock solutions of drugs were
added at the indicated final concentrations in serum deficient media and cells were incubated for the indicated
period of time at 37ºC in 5% CO2. The total number of viable cells remaining post-treatment was determined by
trypan blue exclusion as described above or by crystal violet blue staining assay as described above. Tumor
burden in animals was determined by periodic BLI monitoring. Animals that reached >1X107 p/s/cm2/sr were
enrolled in trials. Erlotinib (Tarceva®) tablets were resuspended in Ora-Plus (Paddock) and administered by
oral gavage.
Luciferase Assays
NIH-3T3 cells and TGF-EGFRWT;InkΔ2/3-/-;PTENlox tumor cell cultures were seeded in triplicates in 6 well
dishes and transfected at a 10:1 ratio using TransIT-LT1 transfection reagent (Mirus) according to
manufacturer’s protocol. 24 hours later, cells were incubated overnight in 0.1% FCS followed by incubation
with vehicle or gefitinib for 24 hours. Lysates were harvested using a dual glow luciferase kit (Promega) and
assayed in a luminometer according to the manufacturer’s protocol. Promoter activity was calculated as the
ratio of firefly luciferase over renilla luciferase and is represented as Relative Light Units.
Flow Cytometry
Intracellular levels of cleaved caspase-3 were measured by flow cytometry. Briefly, vehicle and treated cells
were trypsinized, collected by centrifugation and fixed in 4% formalin for 10 minutes at 37ºC then chilled on ice
for 1 minute. Cells were permeabilized by resuspending in ice-cold 90% methanol for 30 minutes on ice. The
cells were then washed twice with PBS containing 0.5% BSA. A 1:50 dilution of Alexa Fluor 488 conjugated
Cleaved caspase-3 (Asp175) antibody (Cell Signaling Technology) was added to each assay tube and
incubated for 1 hour at room temperature. Cells were washed twice with PBS + 0.5% BSA and resuspended
in 0.5ml PBS. The percentage of Alexa Fluor 488 positive cells was determined by analysis on a Beckman
Coulter CyAn ADP Analyzer (Argon 488nm excitation laser, output measured in the FL1 green fluorescence
channel, logarithmic signal amplification mode). Data was analyzed using Summit software (Beckman Coulter).
Cell cycle profiles were analyzed using a FITC BrdU Flow Kit (BD Pharmingen) according to the
manufacturer’s protocol. Briefly, cells were pulsed for one hour with BrdU at a final concentration of 10 µM in
culture media. Cells were collected by trypsinization and centrifugation then fixed and washed. Cells were
treated with 30 g DNase per tube for one hour at 37ºC to expose incorporated BrdU. Cells were then
incubated with a 1:50 dilution of FITC-labeled anti-BrdU antibody and incubated for 20 minutes at room
temperature. Cells were then washed and resuspended in 20 µl of the DNA content marker, 7-AAD. Levels of
FITC-anti-BrdU (argon laser, FL1 green fluorescence channel, logarithmic signal amplification mode) and 7AAD incorporation (argon laser, FL4 red fluorescence channel, linear signal amplification mode) were
measured on a Beckman Coulter CyAn ADP Analyzer. Data were analyzed using Summit software (Beckman
Coulter).
Quantitative RT-PCR
Total RNA from treated and non-treated cells was extracted with the RNeasy Mini Kit (QIAGEN). cDNA was
prepared with a Superscript® III First-strand Synthesis Supermix for qRT-PCR kit (Invitrogen) according to the
manufacturer’s protocol and PCR was carried out with Taqman probes (TaqMan® Gene Expression Assays,
Applied Biosystems) against 24 gene targets (the names, accession numbers and probe details for these
genes are listed in Table S4). TBP was used as an internal control. Analysis was performed with the
comparative ct method with the comparator being RNA extracted from paired non-treated cells.
Microarray Data Analysis
Total RNA was collected from tumor cells and wild-type cultured astrocytes and from males and female mouse
striatum tissues. RNA was purified using the RNeasy Mini Kit (Qiagen Inc., Valencia, CA, USA) according to
the manufacturer's protocol using 20-30 mg tissue. RNA integrity was assessed using the RNA 6000 Nano
LabChip kit followed by analysis using a Bio- analyzer (Agilent Technologies Inc., Santa Clara, CA, USA). The
sample processing and array work was performed at the MIT BioMicro Center. Affymetrix CEL files were
processed using the Bioconductor suite of programs and custom R scripts (Gentleman et al., 2004). The data
was normalized using RMA protocol and gene expression values calculated for each probeset using GCRMA
library in Bioconductor (Wu and Irizarry, 2004). Linear modeling of expression data incorporating the paired
samples information for the vehicle and gefitinib treatment of the samples were carried out using the limma
package and significantly regulated genes identified using eBayes in limma (Smyth, 2004). The gene
expression values were clustered in the following way to identify the similarity of expression values. One
thousand random genes were identified and their expression values clustered and plotted using the heatmap.2
in Bioconductor. In another set of analysis, only those probesets that showed a standard deviation of greater
than 1.8 units in expression value across the whole data set were retained in the analysis. This resulted in an
expression set of 1002 significantly regulated genes that were analyzed further using the MeV suite of
programs (Saeed et al., 2006). This data was further normalized by mean centering and division by standard
deviation of each gene. A multi-sample (4 samples) Significance Analysis of Microarrays (SAM) analysis
followed by clustering of significantly regulated genes using the default parameters were carried out on this
data set (Tusher et al., 2001) and heatmaps of the 442 significantly clustered genes generated. Finally, TGFEGFRWT;Ink2/3-/-;PTENlox vehicle and gefitinib-treated expression sets were also subjected to GSEA analysis
and the list of top 50 genes that separated the two samples as shown in Figure 6B were used to guide further
experiments (Subramanian et al., 2005).
Merging of Mouse and Human Gene Expression Data
Gene expression data from the mouse models of GBM were Z-score normalized and joined to the unified
TCGA
GBM
data
using
the
MGI
human-mouse
orthology
table
(ftp://ftp.informatics.jax.org/pub/reports/index.html#orthology). From this set, a total of 640 mouse-human
ortholog pairs were identified out of the core 840 genes used to distinguish the variants of human GBM. This
set of 640 mouse genes (core640 set) was summarized to a single value per gene, per tumor group using
averaging. Hierarchical Clustering with p values. This set was subjected to hierarchical clustering using the R
package pvclust. The required R commands are included in the text file “Processing.r” and the input data is the
file “coreAvs.txt”. All the R commands, input data and environment-specific details required to duplicate these
analyses are located here (rous:/home/charliew/n4data/Charest_011510/Array_Data/ForPaper).
Gene set enrichment analysis
GSEA (http://www.broad.mit.edu/gsea/) was used to examine the distribution of the mouse orthologs of the
markers for different human GBM classes in a list of genes ranked based on differential expression between
TGF-EGFRWT;Ink2/3-/- and TGF-EGFRWT;Ink2/3-/-;PTENlox tumor types. All the files required for these
analyses and the results are available here (http://luria.mit.edu/caw_web/Charest_Supplemental/GSEA/).
Hierarchical clustering
Gene expression data from our mouse models of GBM were Z-score normalized and joined to the unified
TCGA
GBM
data
using
the
MGI
human-mouse
orthology
table
found
online
(ftp://ftp.informatics.jax.org/pub/reports/index.html#orthology). From this set, 640 mouse-human ortholog pairs
were identified out of the core 840 genes used to distinguish the variants of human GBM (Verhaak et al.,
2010). This set of 640 mouse genes was summarized to a single value per gene, per tumor group using
averaging.