Liposarcoma Genomic Alterations Define New Targets for Therapy

Samuel Singer, MD Memorial Sloan-Kettering Cancer Center Sarcoma Disease Management Program Chief Gastric and Mixed Tumor Service Liposarcoma • ~20% of 12,000 new STS cases in US each year, the majority are sporadic

• Distinct cytogenetic subgroups comprising 5 subtypes • Simple (translocation-associated) − Myxoid and round cell liposarcoma: >90% t(12;16) • Complex rearrangements (alterations in cell cycle and checkpoint defects) – Well-differentiated and Dedifferentiated liposarcoma: ~90% 12q amplification – Pleomorphic liposarcoma • Genetic events associated with liposarcomagenesis remain undiscovered Therapeutic Challenges • More than 60% of patients with newly diagnosed liposarcoma eventually die of disease

• Diverse histopathology and biological behavior

• WDLS and DDLS are not very responsive to chemotherapy

• Pressing need to develop subtype specific molecularly targeted therapeutics for patients with advanced/recurrent disease Histologic distribution of Liposarcoma subtype MSKCC Clinical Sarcoma Database 1772 liposarcoma patients over 29 years (1982–2011)

Site‐specific histologic subtype distribution

WD Sarcoma Genome Project (Phase I) Genome-wide molecular genetic analysis of 7 sarcoma types

(Phase I)

DNA copy number + LOH 207 sarcomas and

Sequencing 225 matched normals genes and 451 Clinical microRNAs annotation database Transcriptional changes (MSKCC) (-coding genes)

Functional genetic screen of amplified genes in DDLPS Sarcoma Genome Project (Phase I) Analysis of DNA sequence in 6 sarcoma types • Kit in GIST was the most frequently mutated • Next most frequently mutated genes in more than 5% of samples within a subtype were: genes sarcoma type % of samples drug able mutated mutated target PIK3CA myxoid/round cell 18 % PI3K inhibitors liposarcoma TP53 pleomorphic 17 % Small molecules liposarcoma that reactivate mutant p53 (PRIMA-1) NF1 myxofibrosarcoma 10.5 % MEK or mTOR inhibitors pleomorphic 8 % liposarcoma PIK3CA mutations in myxoid/round cell liposarcoma are associated with shorter disease- specific survival compared with wildtype

(p-value = 0.036) E545K helical-domain mutations were associated with increased Akt phosphorylation relative to wildtype in myxoid/round cell liposarcoma

TORC2 phosphorylation site PDK1 phosphorylation site Nucleotide and Copy-number alterations in Dedifferentiated Liposarcoma (n=50) and myxoid/round cell liposarcoma (n=21)

• Consensus plot of statistically significant genome-wide copy-number alterations assessed by RAE Knockdown of the Dedifferentiated Liposarcoma “Amplicome” as a Functional Screen for Potential Driver Genes • Which amplified genes in DDLS are necessary for cancer cell proliferation and survival? – Genomics-driven RNAi screen in 3 genotype-matched DDLS cell lines – Systematic knockdown with shRNAs on 385 significantly amplified genes – 99 genes whose knockdown decreased cell growth, 27 of 99 were amplified in at least one cell line Functional Validation of CDK4 as a Therapeutic Target in DDLS Effect of three validated shRNAs targeting 1m selective CDK4/CDK6 CDK4 on the proliferation of two DDLS cell inhibitor PD0332991 induces a lines G1 growth arrest and senescence Therapeutics for DDLS

• Potential targets: MDM2 antagonists

Apoptosis Following 48hr Nutlin Treatment

No Drug 60 • Nutlin-3 5uM Nutlin – MDM2 50

40

• Roche (R7112) – 30

20

10 prevents MDM2-p53 Annexin Annexin / / % % Apoptosis Apoptosis

0 interaction NADIP DDLS LS141

– CDK4 CDK4/CDK6 inhibitor senescence associated heterochromatic foci • PD0332991 – G1 arrest and senescence

– AURKA AURKA inhibitor DDLS ND DDLS AK (5 µM) • MLN8237 – oral, ATP- competitive, disrupts assembly of mitotic spindle – mitotic delay – …other (IRAK3, PCTK2)? Multi-center phase I trial of R7112 in patients with advanced malignancies • R7112 (Roche) is a small molecule antagonist of MDM2 – binds to the p53 site on the surface of MDM2 and blocks protein-protein interaction between MDM2 and p53 wt protein • Oral administration – MTD: 2500 mg/day, BID x 10 days with 18 days of rest associated with significant thrombocytopenia • Dosing schedule revised to 2500 mg/day, QD dosing x 5 days with 23 days rest • 20 patients with WDLS and DDLS treated to date on BID dosing schedule – 8 with stable disease at 8 weeks • Molecular and pharmacokinetic analysis – p53 mutation, MDM2 levels Phase II clinical trial of PD0332991 in patients with WDLS/DDLS that have CDK4 amplification by FISH and express the Rb protein by IHC • Patients with locally advanced, recurrent or metastatic WDLS or DDLS and disease progression on one prior systemic therapy • PD0332991 dose of 200 mg PO QD for 14 days, followed by 7 days of rest. – safe and tolerable in a phase I study – In phase 1 study two patients with advanced WDLS had stable disease for 3.1 and 2.3 years • Primary endpoint: patients who are progression-free at 12 weeks Phase II clinical trial of MLN8237 in patients with WDLS/DDLS and pleomorphic liposarcoma • Test a second-generation selective AURKA inhibitor in patients with advanced / metastatic liposarcoma

• Primary objective: response rate (CR + PR) assessed at 12 weeks

• Based on a completed phase I study, patients will be treated with MLN8237 50mg PO BID for 7 days, followed by 14 days of rest. Treatment is repeated every 3 weeks (one cycle) Sarcoma Genome Project (Phase II) • Perform a genome-wide genetic and functional analysis of well-differentiated (WDLS) and dedifferentiated (DDLS) Liposarcoma to identify: • distinct genomic subtypes • molecular markers that associate with outcome and pathologic features • genetic alterations associated with sarcoma progression • subtype specific therapeutic targets MSKCC Rockefeller BROAD Samuel Singer Thomas Tuschl Matthew Meyerson Marc Ladanyi Markus Hafner Jordi Barretina Chris Sander Alex Ramos Barry Taylor Shantanu Banerji

Sample procurement and microRNA profiling Systematic shRNA screens selection / RNA and DNA extraction microRNA cloning and sequencing Solexa single molecule cDNA sequencing U133A Affy transcript arrays microRNA ISH 244K Agilent CGH array, Validation of candidate fusion array mutations Agilent microRNA array Computational analysis for protein-coding genes, microRNAs, amplified genes, gene rearrangements, activating mutations and activated pathways microRNA functional analysis Validate genetic, genomic and functional alterations Well-differentiated and Dedifferentiated Liposarcoma

Disease-specific survival for primary Local recurrence-free survival for primary retroperitoneal liposarcoma: WDLS vs. DDLS retroperitoneal liposarcoma: WDLS vs. DDLS Data from MSKCC Sarcoma Database (n=345, 7/1/82 to 6/30/2010)

• Resistant to chemotherapy • Can we exploit differentially expressed microRNAs as therapeutic targets for this deadly disease? Role of miRNAs in liposarcomagenesis • miRNAs regulate cell proliferation, apoptosis, and differentiation • miRNAs are abnormally regulated in cancer – may serve as oncogenes but generally down-regulated in tumors compared to normal tissue – inhibiting miRNA processing enhances tumorigenesis suggest miRNAs act mainly as tumorsuppressors • Profile miRNAs in normal fat (NF), WDLS and DDLS tissue samples to identify miRNAs that associate tissue type and tumor progression – Agilent microarrays (17 NF, 32 WDLS, 30 DDLS) – Deep sequencing of small RNA cDNA libraries (11 NF, 22 WDLS, 22 DDLS) Small RNA cDNA library preparation for sequencing • Developed a set of bar-coded sequencing adapters • parallel sequencing of up to 20 samples in one Solexa sequencing run, from < 2 µg total RNA / sample • > 200,000 sequence reads / sample at a lower cost than miRNA microarrays • map small RNAs to the genome; annotate by functional type • convert cloning frequencies to miRNA expression levels

Hafner, M., Tuschl, T. Methods 2008 44:3-12 Unsupervised clustering of miRNAs in normal fat, WDLS and DDLS samples analyzed by deep sequencing

DDLPS

Normal fat tissue

WDLPS Deep sequencing compared to Agilent microarray (DDLS / NF) miR-21 and and miR-26a are significantly over- expressed in DDLS compared to normal fat

miR-21 miR-26a

PTEN expression Normalized Counts Normalized

NF WD DD NF WD DD NF WD DD miR-26a-2 encoded in the intron of CTDSP2, adjacent to CDK4 on chr12, is amplified in 88% of DDLS miR-143 and miR-145 are down-regulated in WDLS and DDLS compared to NF • miR-143 - most strongly expressed miRNA in normal fat (8% of total) • miR-143 and miR-145 located within 1.8 kb of each other on 5q

miR‐143 miR‐145 Normalized Counts Normalized

N WD DD N WD DD miR‐143F Fold FDR miRF‐145 Fold FDR Change Change

WD/NF ‐3.3 1.5e‐5 WD/NF ‐2.8 > 0.05 DD/NF ‐7.9 4.0e‐12 DD/NF ‐6.6 3.2e‐7 FDR= false discovery rate miR-143 and miR-145 are down-regulated in WDLS and DDLS cell lines compared to ASCs by deep sequencing and quantitative PCR

In situ hybridization for miR-143

DDLS cell lines WDLS cell lines DDLS cell lines WDLS cell lines A lentiviral system was used for stable re-expression of miR-143 and miR-145 in liposarcoma cell lines

miRNA precursor in Tranform E. coli Culture and extract Transfect HEK293T lentiviral vector cells plasmid DNA Cells

miR‐143; miR‐145 re‐ expression Harvest viral Functional supernatant

Studies Select with Puromycin Infect cells of interest (Day 1) (Day 0) Re-expression of miR-143 inhibits proliferation in two DDLS cell lines DNA content (relative to day 2) (relative to day

Time (Days) Time (Days) Re-expression of miR-143, but not miR-145, induces apoptosis in DDLS cells

DDLS14 DDLS8817 1 miR-143 target gene identification in DDLS cells Identify genes with 3’ UTR sequences complementary to the miR-143 but not miR-145 seed sequence Treated 2 DDLS cell lines with miR-143 v. Scr/Untx

Generated triplicate mRNA profiles (Illumina array)

143 upregulated genes 125 downregulated genes

Automated network-based computational approach, NetBox1 1. Cerami E et al., PLoS One, 2010. Network Analysis identified a 24-gene module regulated by miR-143 re-expression

Direct target Downregulated Upregulated Expression of 10 of the 24 genes in the miR-143 network are associated with decreased distant- recurrence–free survival (DFRS) • Expression profiling of 140 primary liposarcoma patients (95 training set, 45 test set) • Expression of 588 genes used to calculate a genomic risk score (GRS) for each patient • In validation: 3-year DRFS of 83% for low GRS vs. 45% for high GRS patients (P=0.001)

The hazard ratio for GRS was 4.4, adjusted for histologic subtype Gobble R et al., Cancer Research, In Press TOP2a, a gene involved in DNA replication, is a predicted direct target of miR-143

DRFS for primary liposarcoma by TOP2A expression

43-fold in DDLS vs. NF

Direct target TOP2A expression Downregulated associated with Upregulated metastasis in DDLS 1 1. Gobble R et al., Cancer Research, In Press Re-expression of miR-143 decreases TOP2a expression and regulates the 3’UTR of TOP2a

DDLS14 DDLS8817 1

DDLS141 DDLS8817 TOP2A is up-regulated in WDLS and DDLS compared to normal Fat Liposarcoma Fold Increase compared False Discover Subtype to Normal Fat Rate Well-differentiated 4.4 6.79E-05 Dedifferentiated 43.0 8.11E-19

Increased expression of TOP2A in DDLS cell lines compared to Adipose Derived Stem Cells (ASC)

ASC LPS141 DDLS8817

TOP2A

α-tubulin

Gobble R et al., Cancer Research, In Press DDLS cell lines TOP2A knockdown inhibits proliferation and induces apoptosis in DDLS cells

DDLS8817 Scr #7 #8

TOP2A

α-tubulin

% apoptosis day 6 after TOP2A Cell proliferation day 6 after TOP2A knockdown knockdown miR-143 re-expression downregulates genes involved in mitosis and cytokinesis: PRC1, , CDC25B, ECT2, and CDC2 (CDK1)

In early anaphase, CDC2 (CDK1) activity falls, PRC1 can now dock to PLK1 and localize it to central spindle for cytokinesis

PRC1 22-fold in DDLS v. NF

In metaphase, CDC25B is recruited to PLK1 and helps Direct target maintain activity of CDC2 (CDK1) Downregulated CDC25B 8-fold in DDLS v. NF Upregulated Neef, R. et al., Nature Cell Biology, 2007 Re‐expression of miR‐143 decreases PRC1 and PLK1 expression and regulates the 3’UTR of PRC1

DDLS141 DDLS8817 DDLS141 DDLS8817

DDLS8817 PLK1 inhibition with BI 2536 decreases proliferation in DDLS cell lines ASC DNA content (relative to Day 0) (relative to Day

DDLS141 DDLS8817 DNA content DNA content (relative to Day 0) (relative to Day (relative to Day 0) (relative to Day PLK1 inhibition induces apoptosis and a G2/M arrest in DDLS cells

DDLS141

DDLS8817

ASC DDLS141DDLS141DDLS8817 Kinases that regulate spindle assembly, mitosis and cytokinesis may be attractive therapeutic targets for DDLS • AURKA is amplified in 30% of DDLS and is a candidate driver gene – Overexpressed 3-fold in DDLS and 10-fold in pleomorphic liposarcoma compared to normal fat – AURKA is prognostic for DRFS in primary liposarcoma (HR=3.1) • PLK1 docking partners PRC1, CDC25B and CDC2 all highly overexpressed in DDLS compared to normal fat – PLK1, PRC1, and CDC2 expression are prognostic for DRFS in primary liposarcoma • Both PLK1 and AURKA interfere with spindle assembly – PLK1 inhibition induces prolonged mitotic arrest and cell death – AURKA inhibition leads to a more retracted mitotic arrest with segregation errors and aneuploidy Summary • Discovered PIK3CA helical and domain mutations in 18% of myxoid / round-cell liposarcoma • Helical PIK3CA mutations are associated with Akt activation in vivo and reduced patient survival • Discovered NF-1 mutations in 10% of myxofibrosarcoma and 8% of pleomorphic liposarcoma • TP53 mutations in 17% of pleomorphic liposarcoma

• These mutations identify a subset of tumors likely to respond to specific pathway inhibitors Summary • miRNA profiles discriminate liposarcoma from normal fat and WDLS from DDLS

• miR-143 is downregulated in WDLS and DDLS compared to normal fat

• Re-expression of miR-143 inhibits proliferation and induces apoptosis in DDLS cells

• miR-143 regulates components of a gene network involved in cell proliferation, apoptosis, and cytokinesis in DDLS cells – Target genes include TOP2a, PRC1, and PLK1

• Targeted inhibition of PLK1 induces apoptosis and cell cycle arrest in DDLS cells Conclusions

• miR-143 downregulation may be an important early event in liposarcomagenesis

• miR-143 may function as a tumor suppressor in liposarcoma and miR-143 re-expression vectors show promise as a multi-targeted miRNA-based therapeutics for WDLS and DDLS

• Selective agents directed at gene products or pathways regulated by miR-143 may have therapeutic value in liposarcoma Acknowledgements Sarcoma Genome Project

Singer Lab Tuschl Lab Meyerson Lab Aimee Crago Markus Hafner Alex Ramos Penelope DeCarolis Sara Hakim Shantanu Banerji Stacy Ugras Rameen Beroukhim Ann Lee Gaddy Getz Christina Angeles Craig Mermel Elliott Brill Wendy Winckler Ryan Gobble Comp Biology Center Barry Taylor RNAi Platform Nick Socci Serena Silver Anders Jacobsen David Root Robert Sheridan Raya Khanin Jordi Barretina Genomics Core Lab Heidi Greulich Agnes Viale Todd Golub Cristina Antonescu Bill Hahn Robert Maki Thomas Tuschl Levi Garraway Gary Schwartz Bill Sellers Marc Ladanyi Eric Lander Chris Sander Matthew Meyerson Harold Varmus MSKCC Rockefeller University BROAD Institute