Cancer Cell, Volume 22 Supplemental Information Genetic Alterations Activating Kinase and Cytokine Receptor Signaling in High-Risk Acute Lymphoblastic Leukemia Kathryn G. Roberts, Ryan D. Morin, Jinghui Zhang, Martin Hirst, Yongjun Zhao, Xiaoping Su, Shann-Ching Chen, Debbie Payne-Turner, Michelle L. Churchman, Richard C. Harvey, Xiang Chen, Corynn Kasap, Chunhua Yan, Jared Becksfort, Richard P. Finney, David T. Teachey, Shannon L. Maude, Kane Tse, Richard Moore, Steven Jones, Karen Mungall, Inanc Birol, Michael N. Edmonson, Ying Hu, Kenneth E. Buetow, I-Ming Chen, William L. Carroll, Lei Wei, Jing Ma, Maria Kleppe, Ross L. Levine, Guillermo Garcia-Manero, Eric Larsen, Neil P. Shah, Meenakshi Devidas, Gregory Reaman, Malcolm Smith, Steven W. Paugh, William E. Evans, Stephan A. Grupp, Sima Jeha, Ching-Hon Pui, Daniela S. Gerhard, James R. Downing, Cheryl L. Willman, Mignon Loh, Stephen P. Hunger, Marco A. Marra, and Charles G. Mullighan Inventory of Supplemental Information Supplemental Data Table S1, related to Table 1 Table S2, related to Table 1 Table S3, related to Table 1 Table S4, related to Table 1 Table S5, related to Table 1 Table S6, related to Table 1 Table S7, related to Table 1 Table S8, related to Table 1 Table S9, related to Table 1 Figure S1, related to Figure 1 Figure S2, related to Figure 2 Figure S3, related to Figure 3 Figure S4, related to Figure 4 Figure S5, related to Figure 5 Figure S6, related to Figure 6 Figure S7, related to Figure 7 Supplemental Experimental Procedures Supplemental References
36
Embed
Supplemental Information Genetic Alterations Activating Kinase … · 2012. 8. 13. · Supplemental Data . Table S1, related to Table 1. List of probe sets that identify Ph-like cases
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Cancer Cell, Volume 22 Supplemental Information
Genetic Alterations Activating Kinase
and Cytokine Receptor Signaling
in High-Risk Acute Lymphoblastic Leukemia Kathryn G. Roberts, Ryan D. Morin, Jinghui Zhang, Martin Hirst, Yongjun Zhao, Xiaoping Su, Shann-Ching Chen, Debbie Payne-Turner, Michelle L. Churchman, Richard C. Harvey, Xiang Chen, Corynn Kasap, Chunhua Yan, Jared Becksfort, Richard P. Finney, David T. Teachey, Shannon L. Maude, Kane Tse, Richard Moore, Steven Jones, Karen Mungall, Inanc Birol, Michael N. Edmonson, Ying Hu, Kenneth E. Buetow, I-Ming Chen, William L. Carroll, Lei Wei, Jing Ma, Maria Kleppe, Ross L. Levine, Guillermo Garcia-Manero, Eric Larsen, Neil P. Shah, Meenakshi Devidas, Gregory Reaman, Malcolm Smith, Steven W. Paugh, William E. Evans, Stephan A. Grupp, Sima Jeha, Ching-Hon Pui, Daniela S. Gerhard, James R. Downing, Cheryl L. Willman, Mignon Loh, Stephen P. Hunger, Marco A. Marra, and Charles G. Mullighan Inventory of Supplemental Information Supplemental Data Table S1, related to Table 1 Table S2, related to Table 1 Table S3, related to Table 1 Table S4, related to Table 1 Table S5, related to Table 1 Table S6, related to Table 1 Table S7, related to Table 1 Table S8, related to Table 1 Table S9, related to Table 1 Figure S1, related to Figure 1 Figure S2, related to Figure 2 Figure S3, related to Figure 3 Figure S4, related to Figure 4 Figure S5, related to Figure 5 Figure S6, related to Figure 6 Figure S7, related to Figure 7 Supplemental Experimental Procedures Supplemental References
Supplemental Data Table S1, related to Table 1. List of probe sets that identify Ph-like cases by PAM. Provided as an Excel file.
Table S2, related to Table 1. limma gene expression signature of Ph+ and Ph-like B-ALL versus non Ph-like B-ALL samples in P9906. Provided as an Excel file.
Table S3, related to Table 1. Validated fusions detected by mRNA-seq Sample ID Support-
mate pairs Gene 1 Gene 2 Comments Rearrangement
Gene Chrom Location Gene Chrom Location
PAKTAL
29 22 5
STRN3 C12orf35 FAM23A
14 12 10
exon 9 intron 4 exon 3
JAK2 AMN1 MRC1
9 12 10
exon 17 intron 6 exon 2
In-frame Aligned to intron In-frame
STRN3-JAK2*,#
C120rf35-AMN1* FAM23A-MRC1*
PAKKCA
66 6 5
EBF1 SEMA6A DOCK8 PAX5 ZCCHC7
5 5 9 9 9
exon 15 exon 1 exon 3-4 exon 6 exon 2
PDGFRB FEM1C CBWD2 ZCCHC7 PAZ5
5 5 2 9 9
exon 11 exon 2 intron 10 exon 3 exon 7
In-frame No ORF Inversion Inversion, disrupted ORF Inversion, disrupted ORF
*detected by deFuse; #detected by Mosaik; **previously identified (Harvey et al., 2010a); ***detected by Trans-ABySS; ORF, open reading frame. Cytokine receptor and kinase-activating fusions highlighted in bold.
Table S4, related to Table 1. Somatic single nucleotide variants (SNVs) and insertion/deletion mutations identified by mRNA-seq Sample ID Gene Chr Position Sequence change Amino acid change PAKHZT JMJD3 chr17 7691628 T>C S433P RAB11B chr19 8373013 G>A A94T JAK2* chr9 5079702 G>A R867Q OFD1 chrX 13696806 T>C S531P;S853P;S993P PAKVKK ILF3 chr19 10655385 G>A R642Q;R646Q NSMAF chr8 59682919 C>T R241H
*, Identified previously (Mullighan et al., 2009c); **, Identified previously (Mullighan et al., 2009b). Multiple SNVs for each gene represent different isoforms.
Table S5, related to Table 1. Validated somatic insertions/deletions and SNVs for PALJDL detected by WGS
Gene Accession Chr Position* Class Sequence change
Amino acid change Comments
IL7R NM_002185.2 chr5 35910329 insertion A>CCCGGGGGTCTGC L242>FPGVC Confirmed by mRNA-seq data
ZNF468 NM_199132 chr19 58036999 frameshift +GGG,-A A67fs Confirmed by mRNA-seq data
CGNL1 NM_032866 chr15 55608184 missense A>T A1027V Low coverage of locus in mRNA-seq
COL4A1 NM_001845 chr13 109628259 missense C>T V883I Low coverage of locus in mRNA-seq
HECA NM_016217 chr6 139529155 missense C>T P105S Confirmed by mRNA-seq data
MOV10 NM_020963 chr1 113042873 missense G>A R841H Confirmed by mRNA-seq data
OR8K1 NM_001002907 chr11 55870928 missense G>C A280P Low coverage of locus in mRNA-seq
PDZD2 NM_178140 chr5 32019253 missense G>A A238T Low coverage of locus in mRNA-seq
PPFIA2 NM_003625 chr12 80212905 missense C>T R922Q Low coverage of locus in mRNA-seq
ROBO1 NM_133631 chr3 78767718 missense C>A G1051C Low coverage of locus in mRNA-seq
VCL NM_014000 chr10 75544616 missense G>A A1071T Confirmed by mRNA-seq data
WDR45L NM_019613 chr17 78167098 missense C>T D341N Confirmed by mRNA-seq data
*aligned to human reference genome 18.
Table S6, related to Table 1. Validated somatic SNVs for PALETF detected by WGS
Gene mRNA_acc Chr Position* Class Sequence change
Amino acid Change Comments
ADNP NM_015339 chr20 48942253 missense G>A S802F Confirmed by mRNA-seq data
ALPK2 NM_052947 chr18 54335283 missense C>T G1926E Low coverage of locus in mRNA-seq
APC NM_001127511 chr5 112204169 missense T>C L1660P Confirmed by mRNA-seq data
ARSI NM_001012301 chr5 149657296 missense C>T V462M Low coverage of locus in mRNA-seq
ATP7B NM_000053 chr13 51437128 missense A>G S584P Low coverage of locus in mRNA-seq
DMPK NM_001081560 chr19 50975027 missense G>A P44L Confirmed by mRNA-seq data
IL7 NM_000880 chr8 79811310 missense C>T G123E Low coverage of locus in mRNA-seq
KRT1 NM_006121 chr12 51356339 missense C>T G488R Low coverage of locus in mRNA-seq
LYSMD4 NM_152449 chr15 98087051 missense C>A V232F Confirmed by mRNA-seq data
MAGI1 NM_015520 chr3 65322046 missense C>T D1196N Low coverage of locus in mRNA-seq
MYOF NM_013451 chr10 95147017 missense T>G K437N Low coverage of locus in mRNA-seq
MYOF NM_013451 chr10 95147027 missense A>T V434E Low coverage of locus in mRNA-seq
PAPPA2 NM_020318 chr1 174830357 missense G>A G332R Low coverage of locus in mRNA-seq
PDE4B NM_001037341 chr1 66156917 missense C>T S31F Low coverage of locus in mRNA-seq
RAD51C NM_058216 chr17 54129161 missense C>A D171E Confirmed by mRNA-seq data
RHAG NM_000324 chr6 49691341 missense C>T E199K Low coverage of locus in mRNA-seq
SPO11 NM_012444 chr20 55342171 missense C>T S81F; S119F Low coverage of locus in mRNA-seq
TSHZ2 NM_173485 chr20 51304885 missense C>T P494L Confirmed by mRNA-seq data
ZNF280A NM_080740 chr22 21199339 missense T>G T206P Low coverage of locus in mRNA-seq *aligned to human reference genome 18.
Table S7, related to Table 1. Somatic deletions for PALJDL and PALETF detected by WGS
CDH2** 15 91176886 NCR 15 91259725 82,839 Deletes exons 1-2 of CDH2 Deletions identified by copy number variant analysis using CONSERTING and CREST that were not detected by SNP array. These algorithms also identified all lesions determined by SNP array analysis. NCR, non-coding region; UTR, untranslated region. *aligned to human reference genome 18; **validated by genomic PCR and Sanger sequencing.
Table S8, related to Table 1. Summary of Ph+ cases and Ph-like prediction by PAM for the COG cohorts P9906, AALL0232 and St Jude Children’s Research Hospital Total XV
P9906
N (%) AALL0232_1
N (%) AALL0232_2
N (%) Total XV
N (%) Total cases 207 283 325 342
Ph+ 0 14 (4.9) 21 (6.5) 7 (2.1)
Ph-like 43 (20.8) 40 (14.1) 42 (13) 33 (9.6)
AALL0232_1 and AALL0232_2 are consecutively enrolled groups from the COG high-risk AALL0232 trial.
Table S9, related to Table 1. Description of COG P9906 and AALL0232 cohorts. Provided as an Excel file. Tabulated information outlining PAM prediction and ROSE clustering determining Ph-like cases, B-cell pathway and kinase activating lesions, and kinase expression (PDGFRB, JAK2, ABL1 and CRLF2) by gene expression profiling using U133 Plus 2.0 array. Column definition is listed below: A: Cohort – P9906 or AALL0232 B: Sample ID - *, mRNA-seq index cases; C: TARGET ID - nomenclature is consistent with previous publications (Mullighan et al., 2009b; Mullighan et al., 2009c; Harvey et al., 2010a) D: BCR-ABL1 status – Positive or negative E: PAM prediction – Ph-like or non Ph-like F: PAM Coefficient > 0.5=Ph-like. Samples are ranked in descending order for both cohorts G: ROSE clustering classification H: Group – TCF3-PBX1, ETV6-RUNX1, BCR-ABL1 positive, MLL rearranged, CRLF2 rearranged or unknown (other) I: B-cell pathway lesions in IKZF1, PAX5, EBF1 or CDKN2A deletion J: CRLF2 - Rearrangements in CRLF2 (IGH@-CRLF2 or PR2Y8-CRLF2) detected previously for P9906 (Harvey et al., 2010a), or over-expression indicating a rearrangement in AALL0232. K: JAK mut: - Mutation in JAK genes detected previously for P9906 (Mullighan et al., 2009c) and identified recently for AALL0232. L: Other kinase-activating lesion: Rearrangements and sequence mutations affecting kinase or cytokine signaling detected by mRNA-seq or whole genome sequencing analysis in the current study. M-R: Gene expression profiling of PDGFRB (M), JAK2 (N-P), ABL1 (Q) and CRLF2 (R)
10
Figure S1, related to Figure 1. Summary of COG ALL recruitment and frequency of Ph-like cases. Approximately two-thirds of COG cases are classified as standard risk, and one third are enrolled onto high-risk trials (P9906, AALL0232). Of the high-risk cases, 15-20% are classified Ph-like, determined by PAM. Within this Ph-like group, 50% harbor rearrangements of CRLF2, with 30% of these expressing concomitant JAK mutations, and the other 20% with unidentified lesions. From the current study, all 15 cases subjected to mRNA-seq harbored rearrangements or mutations affecting kinase and cytokine signaling. Recurrence testing of novel fusions in additional P9906 was prevented by limited availability of RNA. Consequently, the subsequent COG trial of high-risk B-ALL (AALL0232) was used, and the EBF1-PDGFRB fusion was detected in 3/40 (8%) of Ph-like cases. We also identified additional ABL1 and JAK2 rearrangements in this cohort using mRNA-seq. Furthermore, IL7R insertion/deletion mutations were detected in 5/42 (12%) of the P9906 Ph-like cases, but notably, not in AALL0232. No somatic SH2B3 sequence mutations were identified.
11
A
B
C
D
12
E
F
G
H
13
Figure S2, related to Figure 2. Additional fusion validation by RT-PCR RT-PCR and sequencing validation of additional fusions identified by mRNA-seq analysis. Representative RT-PCR gel and sequencing for (A) SEMA6A-FEM1C (S/F) from case PAKKCA; (B) the in-frame fusion TPM4-KLF2 (T/K) from case PAKVKK; (C) OAZ1-KLF2 (O/K) from case PAMDRM; (D) the in-frame fusions FAM23A-MRC1 (F/M) and C12orf35-AMN1 (C/A) from case PAKTAL; (E) the interchromosomal translocation TSHZ2-SLC35A1 (T/S) from case PAKVKK; (F) for ZNF292-SYNCRIP (Z/S) from case PALJDL; (G) the interchromosomal translocation DOCK8-CBWD2 (D/C) from case PAKKCA; and (H) the reciprocal inversion PAX5-ZCCHC7 (P/Z) and ZCCHC7-PAX5 (Z/P), which disrupts the open reading frame of PAX5, from case PAKKCA. NTC = non-template control. (I) Inferred log2 ratio copy number data from SNP array showing gain of one DNA copy between NUP214 and ABL1 in PAKVKK. Each vertical red line indicates log ratio copy number state for a single probe set on the 500K SNP array. (J) Sequencing validation for two additional NUP214-ABL1 cases from the P9906 cohort.
14
A
15
. Figure S3, related to Figure 3. Paired-end mRNA-seq reads aligning to EBF1 and PDGFRB and FISH confirmation of EBF1-PDGFRB in case PAKKCA (A) Paired-end mRNA-seq reads aligning to EBF1 and PDGFRB on chromosome 5q32 for case PAKKCA. The reads aligning to intron 15 of EBF1 and intron 10 of PDGFRB correspond to the genomic breakpoint, and the reads aligning to exon 15 of EBF1 and exon 11 of PDGFRB correspond to the in-frame fusion point. (B)The bacterial artificial chromosome (BAC) clone on chromosome 5 telomeric of EBF1 (RP11-583A20, red). (C) Two BAC clones flanking PDGFRB on chromosome 5; RP11-1079A8 (centromeric, green) and RP11-759G10 (telomeric, red). (D) PDGFRB break-apart assay using probes in B showing loss of the telomeric probe on one chromosome (arrow), due to the deletion between EBF1 and PDGFRB. (E) Colocalization assay showing the fusion signal (arrow) between the telomeric EBF1 clone (red) and centromeric PDGFRB clone (green), Normal signals are close together, but not fused. The fusion was detected in over 95% of cells analyzed, indicating that the EBF1-PDGFRB fusion is present in the predominant clone at diagnosis.
B
C
D
E
16
17
Figure S4, related to Figure 4. Multiple JAK2 breakpoints and genomic mapping of BCR-JAK2 in case PAKYEP. Soft-clipped reads showing two JAK2 breakpoints for the BCR-JAK2 (B/J) fusion in case PAKYEP. (A) Reads aligning to BCR are shown in black letters, with soft-clipped reads detected by CREST aligning to JAK2 (highlighted in blue) and represent either exon 15 (blue underline) or exon 17 (green underline) of JAK2. RT-PCR showing two bands corresponding to JAK2 exon 15 fusion (top band) or JAK2 exon 17 fusion (bottom band). Sequencing validation showing BCR fused to JAK2 exon 15 (left) and exon 17 (right). The red amino acid is changed from the wild-type sequence. (B) Bambino view of mRNA-seq split reads showing the genomic breakpoint of BCR at chr22:21905862 to JAK2 intron 14 (soft-clipped reads) and breakpoint of JAK2 at chr9:5066708 to BCR intron 1 (soft-clipped reads). (C) Genomic PCR and sequencing confirming the breakpoint between BCR and JAK2. Note microhomology of 2bp (CA) at the two breakpoints, which can be aligned to either BCR or JAK2.
B
C
18
Figure S5, related to Figure 5. Genomic PCR and Sanger sequencing validation of additional deletions in PALJDL. (A) Genomic PCR gel. (B) Deletion between non-coding region on chr12:63246783 and RASSF3 (chr12;63291396), which deletes exon 1 of RASSF3. (C) Deletion between ELF1 (chr13:40448351) and non-coding region (chr13:40489525) that deletes exon 1 of ELF1. (D) Deletion between chr7:92281190 and chr7: 92301324 which deletes exon 1 of CDK6. Alignment based on human reference genome 18.
A
B
C
D
19
Figure S6, related to Figure 6. Phosphosignaling analysis in non-Ph-like ALL cases. Primary leukemic cells were thawed, treated with or without dasatinib (100nM) and ruxolitinib (1µM) for 1 hr and stained for pSTAT5 and pCRKL according to Experimental Procedures.
20
Figure S7, related to Figure 7. Modeling EBF1-PDGFRB in vitro. (A) Ba/F3 EBF1-PDGRB cells are sensitive to dasatinib and dovitinib. No cytotoxic effects were observed with cells maintained in factor indicating that imatinib specifically targets the activated PDGFRB and ABL1 kinases. Error bars represent mean + SD of three independent experiments. (B) Imatinib inhibits phosphorylation of the EBF1-PDGFRB fusion protein. (C) pAKT and pERK1/2 are constitutively activated in EBF1-PDGFRB expressing cells, and signaling is inhibited by dasatinib (100 nM).
A
B
C
pAKT pERK
pAKT pERK
21
Supplemental Experimental Procedures
Patients samples and gene expression profiling
Ten Ph-like ALL cases from the COG P9906 high-risk B-ALL study (Bowman et al., 2011), three
cases enrolled on the high-risk COG AALL0232 study (ClinicalTrials.gov Identifier
NCT00075725) and two cases treated on the St Jude Children’s Research Hospital Total XV
(Pui et al., 2009) and Total XVI protocols (ClinicalTrials.gov Identifier NCT00137111 and
NCT00549848, respectively) were selected for mRNA-seq based on a similar gene expression
profile to Ph+ ALL, as determined by ROSE clustering (Harvey et al., 2010b) and PAM
(Tibshirani et al., 2002), and the availability of suitable genomic material. This selection included
a range of cases with variable Ph-like expression signature from strongest to weakest (see PAM
coefficient details in Table S9). Cases were initially chosen from P9906, as the description of
Ph-like ALL was first reported in this cohort (Mullighan et al., 2009b). Details of the P9906
cohort, and prior genomic analyses performed in this cohort, have been described previously
(Mullighan et al., 2009b; Mullighan et al., 2009c; Harvery et al., 20010a; Zhang et al., 2011).
All P9906 patients were classified as high-risk based on the presence of central nervous system
or testicular disease, MLL rearrangement, or based on age, sex, and leukocyte count at
diagnosis. BCR-ABL1 and hypodiploid ALL patients, in addition to those who experienced
primary induction failure were excluded. Cases with high hyperdiploid (as defined by trisomy of
chromosomes 4 and 10 on cytogenetic analysis) or ETV6-RUNX1 cases were excluded unless
central nervous system or testicular involvement was present at diagnosis. A total of 207
enrolled cases had suitable material for 500K SNP microarrays and U133 Plus 2.0 gene
expression microarrays (Affymetrix).
WGS of matched non tumor DNA obtained from remission bone marrow at day 29, or at a
subsequent remission timepoint after commencement of remission-induction therapy, was
performed for all cases. WGS of leukemic cell DNA was also performed for two cases that
22
lacked a chromosomal rearrangement identified by mRNA-seq and prior genomic analyses
(Harvey et al., 2010b). Due to limited availability of RNA material from the P9906 cohort,
recurrence testing of the ABL1, JAK2 and PDGFRB rearrangements was performed in a
separate cohort of B-ALL patients enrolled on the COG AALL0232 study.
All AALL0232 B-ALL patients were diagnosed with National Cancer Institute high-risk ALL
based on WBC count >50x109/L or age >10 years at presentation, prior steroid therapy, or the
presence of testicular disease. The average age at diagnosis was 10.0 + 5.8 years. Twenty-
eight cases (9.9%) were hyperdiploid, 20, (7.1%) were ETV6-RUNX1-positive, 17 (6.0%) were
TCF3-PBX1-positive, 14 (4.9%) were BCR-ABL1-positive, 5 (1.7%) harbored MLL
rearrangements and 199 cases (70.3%) lacked a known chromosomal abnormality. All samples
were obtained with patient or parent/guardian provided informed consent under protocols
approved by the Institutional Review Board at each COG institution. The clinical study was
approved by the National Cancer Institute and appropriate Institutional Review Boards. A total of
231 BCR-ABL1-negative patients had available RNA for RT-PCR.
Recurrence screening for each fusion was also performed on 23 JAK2/MPL-negative MPN
samples from the Harvard myeloproliferative disorders study (age range 35-81), including 13
with polycythemia vera, 5 with essential thrombocythemia and 2 with myelofibrosis (Levine et al.,
2005). In addition, 25 CMML samples obtained from the MD Anderson Cancer Centre (age
range 61-88), and 44 pediatric AML samples (16 cases with normal karyotypes, 18 with
miscellaneous or non-recurrent cytogenetic alterations, 5 inv(16) and 5 t(8;21)) from St Jude
Children’s Research Hospital (Radtke et al., 2009) were also included (age range 3-21).
Sequencing for IL7R and SH2B3 mutations was performed using whole genome amplified
leukemic DNA from the P9906 and AALL0232 high-risk ALL cohorts. All samples were obtained
23
with patient or parent/guardian provided informed consent under protocols approved by the
Institutional Review Board at each COG institution.
Gene expression profiling was performed using U133 Plus 2.0 arrays for P9906 (N=207), and
AALL0232 (N=608) (Affymetrix). Expression signals were normalized by MAS 5.0 algorithm.
Probe sets lacking present calls for every sample were excluded, and signal intensities with
values less than 2 were set to 2. Signals were then log2 transformed for subsequent analysis
(Mullighan et al., 2009b). To identify Ph+ and Ph-like cases, we trained PAM using the second
consecutively recruited subgroup of AALL0232 cases (N=325) to detect Ph+ and Ph-like cases
in the first subgroup of AALL0232 cases (N=283). The PAM predictor containing 257 probe sets
(Table S1) was obtained through cross validation analysis at a threshold of 2.2, and correctly
identified 13 of 14 Ph+ AALL0232_1 cases and classified 40/283 (15%) as Ph-like, determined
by a PAM coefficient greater than 0.5. The same training conditions were applied to the P9906
cohort, with 43/203 cases (21%) classified as Ph-like (Table S9). To identify differentially
expressed genes in Ph-like cases, limma (Linear Models for Microarray Analysis) (Smyth, 2004)
with estimation of fold-change and false discovery rate was also performed (Benjamini and
Hochberg, 1995) (Table S2).
mRNA-seq library preparation and sequencing
mRNA-seq was performed as previously described (Morin et al., 2010) with modifications. Total
RNA was extracted from leukemic cells obtained from bone marrow aspirates or peripherial
blood using TRIzol (Life Technologies). Poly(A)+ RNA was enriched from 5-10 µg of DNAse 1-
treated total RNA using the MACS mRNA isolation kit (Miltenyi Biotec). Double-stranded cDNA
was synthesized from the purified poly(A)+ RNA using the Superscript Double-Stranded cDNA
Synthesis kit (Life Technologies) and random hexamer primers (Life Technologies) at a
concentration of 5 μM. The cDNA was fragmented by sonication and a paired-end sequencing
library prepared following the paired-end library preparation protocol (Illumina). For mRNA-seq
24
library sequencing, clusters were generated on the Illumina cluster station and paired-end
sequence reads were generated using v3-v5 sequencing reagents on the Illumina GAIIx and
HiSeq 2000 platforms following the manufacturer's instructions. Read length summary for each
case is provided in below. Image analysis, base-calling and error calibration were performed
using v1.0, v1.3.2, v1.5.0 and v1.6.0 of Illumina's Genome analysis pipeline.
Read summary information for mRNA-seq data and matched constitutional DNA
*, samples sequenced on the HiSeq 2000.
Sample ID Sample type Number of lanes x read length (bp) Total reads