Sarcomas with Aberrant Transcription Factors: Biology and Expression Profiling Marc Ladanyi Memorial Sloan-Kettering Cancer Center New York, NY, USA.

Sarcomas with Aberrant Transcription Factors: Biology and Expression Profiling

Marc Ladanyi

Memorial Sloan-Kettering Cancer CenterNew York, NY, USA

Translocation-associated sarcomas

1. General biological features and comparison to sarcomas with non-specific cytogenetic alterations

2. Insights from microarray-based expression profiling of translocation-associated sarcomas

Prom oter Substitution

Chim ericTranscription Factor

Chim ericTyrosine Kinase

Gene Fusion

Balanced

Loss o f Tum or Suppressor Gain of Oncogene

Unbalanced

Translocations

Pathologic genetic rearrangements in human cancers: a family tree

Deregulated gene expression

Deregulated growth signaling

Molecular pathology of sarcomas: two major classes

• approx. 1/3 of all sarcomas• 15 different sarcoma types

with over 25 different translocations

• fusion genes: aberrant chimeric transcription factors (most) or aberrant kinases (some)

• biology: transcriptional deregulation or aberrant signaling

1. Sarcomas with specific reciprocal translocations and relatively simple karyotypes

• approx. 2/3 of all sarcomas• biology: genetic gains &

losses, chromosomal instability, telomere dysfunction

2. Sarcomas with complex unbalanced karyotypes

and no specific translocations

Sarcomas with specific

translocations

Sarcomas with non-specific genetic

alterations

Karyotypes Usually simple Usually complex

Translocations Reciprocal & specific, producing fusion genes

Non-reciprocal & non-specific, causing gene copy number changes

Telomere maintenance mechanisms

Telomerase common, ALT mechanism rare

ALT mechanism more common than telomerase

P53 pathway alterations

Relatively rare, but strong prognostic impact*

More frequent, but limited or no prognostic impact

Incidence in P53-mutant or knockout mice

Not observed Common

Incidence in bilateral retinoblastoma and Li-Fraumeni syndrome

Rare Common

Global gene expression profiles

Robust clustering Looser clustering


* MSKCC study: Genes Chromos Cancer 2004

Evidence of “alternative lengthening of telomeres”

(ALT) telomere maintenance mechanism:

Ewing sarcoma (0/30) vs osteosarcoma (38/60)

(P<0.0001)*


translocations


alterations








Relatively rare, but strong prognostic impact*



Not observed Common


Rare Common




* Hopkins study: AJP 2004

Evidence of “alternative lengthening of telomeres”

(ALT) telomere maintenance mechanism:

Translocation sarcomas (0/9) vs other sarcomas (7/9)

(P=0.002)*


translocations


alterations








Relatively rare, but strong prognostic impact



Not observed Common


Rare Common




* MSKCC study: JCO in press0 20 40 60 80 100 120

0.0

0.2

0.4

0.6

0.8

1.0

P53 mutated (n=8)

No P53 mutation (n=52)

P53 in Ewing sarcoma*

p<0.0001


translocations


alterations








Relatively rare, but strong prognostic impact



Not observed Common


Rare Common




Major fusion genes in sarcomas: biological overview

Sarcoma type Translocation Fusion gene Transcriptional Deregulation

Aberrant Signaling

Ewing sarcoma t(11;22) EWS-FLI1 X

t(21;22) EWS-ERG X

Clear cell sarc. t(12;22) EWS-ATF1 X

Myxoid LPS t(12;16) TLS-CHOP X

Alveolar rhabdomyo-sarcoma

t(2;13) PAX3-FKHR X

t(1;13) PAX7-FKHR X

DSRCT t(11;22) EWS-WT1 X

Extr. myx. CS t(9;22) EWS-CHN X

Synovial sarc. t(X;18) SYT-SSX1,2 X

DFSP t(17;22) COL1A1-PDGFB X

Cong. FS t(12;15) ETV6-NTRK3 X

IMT t(2p23) ALK fusions X

End. str. sarc. t(7;17) JAZF1-JJAZ1 X

ASPS t(X;17) ASPL-TFE3 X

Low grade MFS t(7;16) FUS-BBF2H7 X

Pericytoma t(7;12) ACTB-GLI X

• Slow progress in identifying genuine biologically critical target genes of chimeric transcription factors – e.g. well-established targets:

• EWS-FLI1 <15• EWS-WT1, PAX3-FKHR <5 each

– Low throughput gene-by-gene studies– False leads generated by assays using exogenous target

promoters and/or heterologous cells– Most representative cellular background for inducible

systems still subject of active investigation– Need for more alternative higher throughput approaches

Target Genes of Chimeric Transcription Factors in Sarcomas

Target Genes of Chimeric Transcription Factors in Selected Sarcomas

Tumor Fusion protein Type of DNA BD

Direct or indirect targets

Ewing sarcoma / PNET

EWS-FLI1EWS-ERG

ETS TGFBR2 (), p57KIP2 (), MYC, PDGF-C, ID2, CCND1, UBE2C, IGFBP3()

DSRCT EWS-WT1 Zn finger PDGF-A, IL-2/15R, TALLA1, BAIAP3

Alveolar RMS PAX3-FKHRPAX7-FKHR

Paired box MET, CXCR4

Synovial Sarcoma

SYT-SSX1SYT-SSX2

(none) XRCC4, TLE1

Alveolar Soft Part Sarcoma

ASPL-TFE3 bHLH-LZ CYP17A1, UPP1

Sarcomas with Aberrant Transcription Factors

1. General biological features and comparison to sarcomas with non-specific cytogenetic alterations

2. Insights from microarray-based expression profiling of translocation-associated sarcomas

Expression Profiling of Sarcomas with Chimeric Transcription Factors

Why is it interesting?

1. Translocation-associated sarcomas already have an objective molecular classification (detection of specific translocation) that can be used to evaluate unsupervised clustering based on expression profiles

2. Transcriptional deregulation is likely to be central to the pathogenesis of translocation-associated sarcomas with chimeric transcription factors

Alveolar Rhabdomyosarcoma(ARMS)

23 16 PAX3-FKHR 7 PAX7-FKHR

Desmoplastic Small Round Cell Tumor (DSRCT)

32 32 EWS-WT1

Synovial Sarcoma (SS) 46 25 SYT-SSX121 SYT-SSX2

Ewing Sarcoma/PNET (ES)

38 22 EWS-FLI1 type 111 EWS-FLI1 type 2 5 EWS-ERG

Alveolar Soft Part Sarcoma(ASPS)

14 11 ASPL-TFE3 type 1 3 ASPL-TFE3 type 2

• 137 tumor samples from MSKCC and U.Penn. (F. Barr)• + 4 xenografts + 12 cell lines = 153 total samples• Classification of all samples confirmed by fusion transcript RT-PCR

Expression profiling of sarcomas with chimeric transcription factors

Gene expression data analysis

• Hybridization to Affymetrix U133A GeneChip (22215 probe sets, 18500 transcripts, 14500 genes)

• Unsupervised Hierarchical Clustering

• Selection of differentially expressed genes Two-tailed t-tests; p<0.01 after Bonferroni correction

Raw data processing Clustering Number of genes

MAS v5.0 Cluster/Treeview Subsets

RMA method Pearson correlationAll probe sets (22215)

Unsupervised hierarchical clustering n°1MAS v5.0, Cluster/Treeview, different subsets of most variable probe sets

ARMS SS ES DSRCT SS ES ASPS

7200 probe sets

2200 probe sets

3200 probe sets

ASPS DSRCT ARMS ES SS ES SS

DSRCT ASPS SS ARMS SS ES ARMS

Variability of clustering results according to number of probe sets selected

Unsupervised Hierarchical Clustering n°2

RMA method, Pearson correlation, all 22215 probe sets

ES ASPS DSRCT ARMS ES SS

n=153

celllines

Unsupervised hierarchical clustering n°2

RMA method, Pearson correlation, all 22215 probe sets

ASPS ES SS DSRCT ARMS n=137tumors

only

ES/PNET orphan

case

Ewing’s sarcoma

Alveolar rhabdomyosarcoma

Alveolar soft part sarcoma

Synovial sarcoma

Desmoplastic small round cell tumor

Unsupervised hierarchical clustering Multi-dimensional scaling analysis

n=137

3 different views of same data

Distribution of tumor types and numbers of differentially expressed genes by sarcoma type

Numbers of differentially expressed genes “inflated” by comparison to reference group of remaining four sarcomas, each with strong distinctive expression profile

Samples

Genes significantly over- or under-expressed*

Subset with 2 fold

overexpression*

SS 46 6816 638

ARMS 23 1518 282

DSRCT 28 3163 554

ASPS 12 1590 531

ES/PNET

28 2157 294

* all significant at Bonferroni p<0.01 relative to 4 other sarcoma types

The gene expression profiles of translocation sarcomas contain many previously reported differentially expressed genes

“Literature validation”

Sarcoma Gene P-value Fold change

ARMS MyoD1 10-18 12

FGFR4 10-11 7.7

Myogenin 10-9 4.5

ES MIC2 10-18 2.8

MYC 10-17 5.2

CCND1 10-12 4.5

SS PRAME 10-39 5.6

TLE1 10-33 8.5

FZD1 10-24 8.4

• Differentially expressed genes in sarcomas with chimeric transcription factors– Target genes regulated by fusion protein (direct or indirect) – Genes reflecting pre-existing phenotype of host cell– Genes reflecting secondary genetic or epigenetic alterations

Use of microarray data to predict classification of translocation sarcomas

Multiple-class Prediction using Supervised Methods Support Vector Machine Prediction Rule

Validation set

Creation of a Prediction Model

Predictor/Classifier

Selection of a subset of informative genes for each class

10 fold Cross

validation

Training set

90% 10%

Assess accuracy on the validation set

Performance of microarray-based predictor

136/137 sarcomas were accurately predicted

23 0 0 0 0

0 12 0 0 0

0 0 28 0 0

0 0 0 46 0

1 0 0 0 27

ARM

SAS

PSD

SRC

TSS ES

ARMS

ASPS

DSRCT

SS

ES

Model Prediction

Histological and

Molecular Diagnosis

Similar results with other predictions methods Single misclassified case = orphan case in unsupervised clustering

ES with EWS-FLI1 type 1 fusion + p16 deletion and no PAX-FKHR

– Global analyses / Microarray-based classifier• Translocation sarcomas are associated with very

distinctive gene expression profiles – can be used to classify these sarcomas as accurately

as translocation detection

• Parameters for raw data processing and clustering can have strong effects on unsupervised analyses

– Contribution of chimeric transcription factor target genes to specific expression profiles


Contribution of chimeric transcription factor target genes to specific expression profiles of translocation sarcomas

Approaches

1. Differential expression of known target genes2. Cross-referencing of profiles from microarray

experiments using inducible cell lines3. Identification of candidates for target gene

analyses based on expression profiling data from primary tumors

The expression profiles of translocation sarcomas contain known target genes of their chimeric transcription factors

ES: EWS-FLI1 targets

ProposedTarget Gene

Reported Effect of EWS-FLI1

Differential expression in ES/PNET (n=38 vs 115)

P-value Fold change

MYC 10-17 5.2ID2 10-9 3.2

CCND1 10-12 4.5UBE2C 10-8 2.7PDGFC 10-9 3.5

The expression profiles of translocation sarcomas contain known target genes of their chimeric transcription factors

DSRCT: EWS-WT1 targets

Proposed Target Gene

Reported Effect of

EWS-WT1

Differential expression in DSRCT (n=32 vs 121)

P-value Fold change

TALLA-1 10-15 5.5

PDGFA 10-14 2.7

IL2RB 10-13 4.6

BAIAP3 10-9 4

EWS-WT1 target genes defined in a heterologous cell line are over-represented among genes

differentially expressed in DSRCTs

• Induction of EWS-WT1 protein expression in U2OS human osteosarcoma cells with tetracycline-inducible EWS-WT1 construct

• Hybridized to U133A chips• W. Gerald Lab, MSKCC

• 102 genes demonstrated at least a 3 fold alteration in expression level at 24h following induction of EWS-WT1 – 22 down-regulated, 80 upregulated



• 17-fold enrichment for EWS-WT1 target genes among genes in the DSRCT expression profile (Chi-square p<0.0001)

• include several previously validated EWS-WT1 targets: – BAIAP3, PDGFA, TALLA1, IL2RB

80 genes upregulated at least 3 fold

553 genes that were 2 fold overexpressed in DSRCT relative to other translocation

sarcomas

U2OS cell experiment Expression profiles of DSRCT tumors

35 genes in common



• 44% of genes upregulated by induction of EWS-WT1 in the U2OS human osteosarcoma cell line were also significantly overexpressed in DSRCTs

• 6% of the DSRCT expression profile corresponds to genes induced by EWS-WT1 in the model system



• 44% of genes upregulated by induction of EWS-WT1 in the U2OS human osteosarcoma cell line were also significantly overexpressed in DSRCTs

• 6% of the DSRCT expression profile corresponds to genes induced by EWS-WT1 in the model system

• Comparison with similar data in Ewing’s sarcoma• Lessnick SL, Dacwag CS, Golub TR. Cancer Cell 1:393-401, 2002

• 46% of the EWS-FLI1-upregulated genes appeared in the ES/PNET expression profile obtained from primary tumors

• 8% of the ES/PNET expression profile corresponded to genes induced by EWS-FLI1 in the model system

Alveolar Soft Part SarcomaTop 20 significantly overexpressed genes

Fold change P-value Gene

121 10-11 CYP17A1

86 10-24 INHBE

75 10-9 MIBP

62 10-37 GPNMB

52 10-28 GOS2

51 10-34 Hs.57548

49 10-9 DEFB1

47 10-11 SULT1C1

47 10-10 UPP1

43 10-11 SULT1C1

43 10-15 SV2B

42 10-9 GPR56

41 10-7 NTSR2

40 10-19 Hs.37189

36 10-7 AGXT2L1

30 10-14 CLI

29 10-16 CLI

27 10-2 3 PTDGS

25 10-29 PTDGS

24 10-39 PTDGS

Ranked by fold-change

Can we use this list to identify new target genes of sarcoma

fusion proteins?

Makoto Nagai, MD PhD



121 10-11 CYP17A1

86 10-24 INHBE

75 10-9 MIBP

62 10-37 GPNMB

52 10-28 GOS2

51 10-34 Hs.57548

49 10-9 DEFB1

47 10-11 SULT1C1

47 10-10 UPP1

43 10-11 SULT1C1

43 10-15 SV2B

42 10-9 GPR56

41 10-7 NTSR2

40 10-19 Hs.37189

36 10-7 AGXT2L1

30 10-14 CLI

29 10-16 CLI

27 10-2 3 PTDGS

25 10-29 PTDGS

24 10-39 PTDGS


0

5000

10000

15000

20000

25000

30000

35000

40000

45000

1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151

Cytochrome P450 subfamily 17A1 (CYP17A1)

• Promoter region of CYP17A1 contains 3 potential TFE3 sites

• CYP17A1 promoter strongly activated by ASPL-TFE3, but not native TFE3

• Strong physical binding of ASPL-TFE3 to sites #1 and #3 by EMSA

• In vivo presence of ASPL-TFE3 at CYP17A1 promoter in model cell line (293) by chromatin IP

• Induction of ASPL-TFE3 in model cell line (293) results in upregulation of endogenous CYP17A1 by real-time Q-RT-PCR

Identification of CYP17A1 as a direct target of ASPL-TFE3 based on its strong differential overexpression

in the expression profile of ASPS



121 10-11 CYP17A1

86 10-24 INHBE

75 10-9 MIBP

62 10-37 GPNMB

52 10-28 GOS2

51 10-34 Hs.57548

49 10-9 DEFB1

47 10-11 SULT1C1

47 10-10 UPP1

43 10-11 SULT1C1

43 10-15 SV2B

42 10-9 GPR56

41 10-7 NTSR2

40 10-19 Hs.37189

36 10-7 AGXT2L1

30 10-14 CLI

29 10-16 CLI

27 10-2 3 PTDGS

25 10-29 PTDGS

24 10-39 PTDGS


Can we do this with another gene from the list?



121 10-11 CYP17A1

86 10-24 INHBE

75 10-9 MIBP

62 10-37 GPNMB

52 10-28 GOS2

51 10-34 Hs.57548

49 10-9 DEFB1

47 10-11 SULT1C1

47 10-10 UPP1

43 10-11 SULT1C1

43 10-15 SV2B

42 10-9 GPR56

41 10-7 NTSR2

40 10-19 Hs.37189

36 10-7 AGXT2L1

30 10-14 CLI

29 10-16 CLI

27 10-2 3 PTDGS

25 10-29 PTDGS

24 10-39 PTDGS


203234_at gb:NM_003364.1 Uridine phosphorylase (UP)

02000400060008000

1000012000140001600018000

1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151

Uridine Phosphorylase (UPP1)

• UPP1 promoter more strongly activated by ASPL-TFE3 than by native TFE3

• Induction of ASPL-TFE3 in model cell line (293) results in upregulation of endogenous UPP1, as measured by real-time quantitative RT-PCR

Identification of UPP1 as a direct target of ASPL-TFE3 based on its strong differential overexpression


Identification of UPP1 as a direct target of ASPL-TFE3 based on its strong differential overexpression




Identification of UPP1 as a direct target of ASPL-TFE3 based on its strong differential

overexpression in the expression profile of ASPS



• Potential therapeutic interest of uridine phosphorylase – converts the pyrimidine

analogue, 5'-deoxy-5'fluorouridine, to 5-FU

– allows administration of 5'-deoxy-5'fluorouridine as a prodrug with low toxicity to non-neoplastic cells expressing only basal levels of uridine phosphorylase

Identification of potential target genes of SYT-SSX based on strong differential overexpression in the

expression profile of synovial sarcoma

Top 5 ranked by p-value

TLE1 (transducin-like enhancer of split 1) a transcriptional repressor of

Wnt/-catenin signaling

Tsuyoshi Saito, M.D. Ph.D.

See Pathology Poster # 5(not in printed program)

– Global analyses / Microarray-based classifier– Contribution of chimeric transcription factor

target genes to specific expression profiles• Significant subsets of genes in specific expression

profiles may be chimeric transcription factor target genes

• Specific expression profiles can be used to identify new candidates for target gene analyses


Kinases as therapeutic targets in sarcomas with chimeric transcription factors

• kinases or signaling pathways directly activated by the specific aberrant transcription factor

Tumor Fusion protein Signaling proteins confirmed or proposed to be upregulated

Ewing sarcoma /PNET

EWS-FLI1EWS-ERG

PIM3

Alveolar RMS PAX3-FKHRPAX7-FKHR

MET

Desmoplastic small round cell tumor

EWS-WT1 PDGF-A, FGFR4

Kinases as therapeutic targets in sarcomas with chimeric transcription factors

• kinases or signaling pathways directly activated by the specific aberrant transcription factor

• kinases or signaling pathways overexpressed in specific sarcomas apparently unrelated to direct action of aberrant transcription factor

• ERBB2/Her2/neu in synovial sarcoma

• EGFR in synovial sarcoma

• KIT in Ewing’s sarcoma• kinases activated by mutations as secondary or

cooperating events in sarcomas with aberrant transcription factors (like FLT3 mutations in leukemias)

• none identified so far• chimeric transcription factors regulated by

phosphorylation • EWS-WT1

MSKCC • Ladanyi Lab

– Tsuyoshi Saito – Makoto Nagai – Marick Laé– Violetta Barbashina– Man Yee Lui– Zhiquan Zhao

• Biostatistics– Adam Olshen– Shannon Chuai

• W. Gerald Lab– William Gerald– Lishi Chen

• Other MSKCC collaborators– Cristina Antonescu– John Healey, Murray Brennan, Sam Singer– Paul Meyers, Len Wexler, Robert Maki

• Johns Hopkins– Pete Argani

Expression profiling of sarcomas with aberrant transcription factors and related studies

• U. Pennsylvania – Fred Barr

• U. Michigan – Larry Baker– Dafydd Thomas

Sarcomas with Aberrant Transcription Factors: Biology and Expression Profiling Marc Ladanyi Memorial Sloan-Kettering Cancer Center New York, NY, USA.

Documents

major classessarcomas

prognostic impactincidence

bilateral retinoblastoma

knockout mice

major classesapprox

aberrant transcription

aberrant signaling1

genetic gains losses