A Functional Genomics Approach to Autophagic Cell Death Gene Discovery Genome Sciences Centre British Columbia Cancer Agency CATGGCGTGGGGAT CATGGCTAATAAAT CATGGCTCAAGGAG CATGGCTGGACTCC CATGGCTGTGGCCA CATGGCTTTCGTGT CATGGCTTTTTGGC CATGGGAACCGACA CATGGGACCGCCCC CATGGGACCGCTCA CATGGGATCACAAT CATGGGCAACGATC CATGGGCAGCAAGC CATGGGCAGCAATT

Acknowledgements BC Cancer Agency BC Cancer Foundation National Cancer Institute of Canada Michael Smith Foundation for Health Research NSERC GSC PCD group Sharon Gorski Suganthi Chittaranjan Doug Freeman Melissa McConechy Jennifer Kouwenberg Bioinformatics Steven Jones Erin Pleasance Richard Varhol Scott Zuyderduyn GSC Sequencing Group University of Maryland Biotech Institute Eric Baehrecke

Outline Programmed Cell Death (PCD) A genomic approach to gene identification in Drosophila PCD Validation of candidate Drosophila PCD genes and an RNAi screen to assess function.

Programmed cell death (PCD) PCD is a genetically regulated type of cell death in which the cell uses specialized cellular machinery to kill itself; it is a cell suicide mechanism that enables metazoans to control cell number and eliminate cells that threaten the animal's survival Types (Schweichel & Merker, 1973): Type I = apoptosis Type II = autophagic cell death Type III = non-lysosomal

J. Mol. Recognit. 2003; 16: 337–348 Aims Molecular machinery involved? Relationships? Which genes are necessary & sufficient? Which genes are associated with human disease?

Genome Sciences Centre Programmed Cell Death Group Apoptotic Cell DeathAutophagic Cell DeathAutophagy inxs (Doug Freeman) echinus (Ian Bosdet) Cloning and Characterization Mammalian cell line transcription profiling and RNAi (M. Qadir) Gene expression profiling (SAGE) of autophagic PCD in Drosophila salivary glands Role of Akap200 (Claire Hou) Role of CG4091 (Suganthi Chittaranjan) Bioinformatic analyses: associations between autophagic PCD, apoptosis, autophagy, and cancer (Erin Pleasance) Novel Gene Discovery (Brent Mansfield) RNAi screen in Drosophila cell line (Suganthi, Melissa McConechy, Jennifer Kouwenberg, Amy Leung) RNAi screen in mammalian cell line (M. Qadir)

Types of Programmed Cell Death (PCD) (adapted from Baehrecke, 2002) I. Apoptosis II. Autophagic PCD

Distinctions between Type I & II PCD CharacteristicType I (Apoptosis)Type II (Autophagic) HallmarkCondensation, membrane blebbing & apoptotic body formation Formation of autophagic vacuoles Typical OccurrenceIsolated cellsGroups of cells CytoskeletonCytoskeletal collapseCytoskeletal preservation CytoplasmCondensation & fragmentation Engulfed by autophagic vacuoles OrganellesPreservedEngulfed by autophagic vacuoles NucleusChromatin condensation & nuclear fragmentation early Degradation is late; follows cytoplasmic degradation DegradationPhagocytes or neighbouring cells (heterophagy) Autophagic vacuoles fuse with lysosomes (autophagy); remnants by phagocytes CaspasesCaspase-dependentCaspase-independent or - dependent TriggersDNA damage, oncogene activation, extracellular signals, etc. Ras activation, ecdysone, constitutive autophagy, ??

Autophagy Autophagosome (Double membrane) Autophagolysosome The Cell, A Molecular Approach, G.M. Cooper, Ed., 2000 Housekeeping: low level Starvation: upregulation, provides nutrients PCD: autophagy also upregulated. Paradox?

Autophagic PCD in Development Dictyostelium sorocarp formation insect metamorphosis intersegmental muscle, gut, salivary glands mammalian embryogenesis regression of interdigital webs, sexual anlagen mammalian adulthood intestine, mammary gland post-weaning, ovarian atretic follicles

Autophagic PCD in disease & disease models Neurodegenerative diseases (Alzheimers, Parkinson, Huntington’s, Lurcher mouse) cardiomyocyte degeneration spontaneous regression of human neuroblastoma tamoxifen-treated mammary carcinoma cells (MCF-7) TNFα-treated T lymphoblastic leukemic cells bcl-2 antisense treatment of human leukemic HL60 cells Oncogenic Ras-expressing human glioma and gastric cancer cells beclin-1 is an autophagy gene that is monoallelically deleted and expressed at reduced levels in human breast and ovarian cancers; beclin-1 knockout mouse indicated that beclin-1 is a haploinsufficient tumor suppressor gene; hets display an increased incidence of lymphoma, lung carcinoma and liver carcinoma

Experimental Approach Gene expression profiling (SAGE) and RNAi: Comprehensive Gene Discovery Drosophila model system: Known cell death genes/pathways are conserved Genetic and molecular tools Sequence resources FlyBase and GadFly databases Multiple tissues undergo PCD; well-characterized morphologically

The Drosophila Salivary Glands Cell types: duct cells & secretory cells Cell number: ≈ 100 cells/gland Size of gland: ≈ 150 x 900 µm Total RNA/pair of glands: ≈ 0.6 µg (20 pairs/microSAGE library; 500 pairs /cDNA library) Development: ectodermally-derived during late embryogenesis; during metamorphosis, a pulse of ecdysone triggers larval salivary gland PCD; adult salivary glands arise from a pair of imaginal rings (from Kucharova-Mahmood et al., 2002)

Drosophila salivary gland PCD (adapted from Jiang et al., 1997) autophagicstage-specificsynchronous 20 hr24 hr26 hr known cell death genes are highly conserved and regulated transcriptionally hr (APF, 18°C) RT diap2 rpr hid

ESTs 3’ ESTs from salivary gland specific cDNA library 500 pairs of salivary glands from mixed stages, 16-24hrs) High quality 3’ ESTs5181 Number of different transcripts represented1696 Matches to BDGP* predicted genes AND BDGP ESTs1280 Matches to BDGP ESTs only (but no predicted gene) 145 Matches to BDGP predicted genes only (no other ESTs) 75 No matches to BDGP predicted genes or ESTs 196 *Berkeley Drosophila Genome Project Release 2

ESTs Cluster size Top 5 most abundant salivary gland ESTs: MT35 (mitochondrial large rRNA)733 CG4151 (no annotation) 375 Eig71Ec (Ecdysone-induced gene 71Ec) 171 CG3132 (beta-galactosidase) 155 CG14062 (DNA/RNA non-specific endonuclease) 74 3’ ESTs Number of clusters Cluster Size Distribution

SAGE (Velculescu et al. 1995) Potential for gene discovery

Salivary gland SAGE: Tag mapping summary (S. Gorski et al., Curr Biol 13: , 2003) (E. Pleasance et al., Genome Res 13: , 2003) SAGE Library Tags analyzed Transcripts Total transcripts 16 hr34,9893,126 4, hr31,2153, hr30,8232, % 6.2% 6.5% 25.3% known or predicted genes genomic DNA and EST (but no annotated gene) genomic DNA only no match

1244 transcripts are expressed differentially (p<.05) prior to salivary gland PCD 512 genes have associated biological annotations (Gene Ontology in Flybase) 732 genes have unknown functions 377 of these genes were not predicted (GadFly Release 2) 48 correspond solely to salivary gland ESTs

SAGE Identifies Genes Associated Previously With Salivary Gland Death Tag Frequency BFTZ-F1 EcR/USP BR-C E74 E93 rpr hid ark dronc crq iap2 Cell Death E75

Genes associated with autophagic PCD Expression fold- difference (16 hr vs 23 hr) Protein synthesis Hormone related Trans- cription* Signal transduction Apoptosis Immune response/ TNF-related Autophagy Unknowns

Gene expression is reduced in a salivary gland death-defective mutant Fold-difference in expression (16 hr vs 23 hr) E93 is an ecdysone-induced gene that encodes a DNA binding protein required for salivary gland cell death (Lee et al., 2000, 2001) Genes with map locations corresponding to E93 binding sites and upregulated prior to salivary gland PCD were tested by QRT-PCR:

Function-based strategies for characterizing differentially expressed genes Mutants available Phenotype analyses salivary glands, midguts, retinas, embryos Overexpression and loss-of-function in vivo Prioritization midgut PCD human ortholog/cancer l(2)mbn cells siRNA in mammalian cells RNAi in Drosophila l(2)mbn cells Mutants unavailable

Prioritization Differentially expressed (p < 0.05) tags that unambiguously correspond to known/predicted genes and show at least 5-fold difference in expression (= 489) similar differential expression prior to midgut PCD (Li & White, Dev Cell, 2003, & in-house QRT-PCR) (> 182; in progress) mammalian ortholog (53%; InParanoid, Remm et al., 2001) mammalian ortholog differentially expressed in cancer (in progress) present in RNAi cell system (64% by Affymetrix analysis of l(2)mbn cells)

Finding PCD genes by orthology and expression Human cancer and normal SAGE libraries (102 libraries from CGAP) Differentially expressed genes (p<.05 = 2277 genes) Drosophila SAGE libraries Differentially expressed genes (16 hr vs 23 hr, p<.05 = 564 genes) Human orthologues (296/564 have human RefSeq ortholog) Set of Drosophila/human orthologues perturbed in both cancer and Drosophila PCD (= 23 Drosophila genes) E. Pleasance, S. Gorski and S. Jones

Genes upregulated in PCD E. Pleasance

CG4091 Met all criteria: Upregulated prior to salivary gland PCD (X 105 in SAGE) Upregulated prior to midgut PCD (X 9) Expressed in mbn2 cells Human ortholog (TNF-induced protein GG2-1/SCC-S2) GG2-1/SCC-S2 possibly associated with human cancer: SCC-S2 amplified in a metastatic head and neck carcinoma-derived cell line compared to matched primary tumor-derived cell line (Kumar et al., JBC, 2000) A “Double knock-out” strategy Mbn2 cell line Addiap1 dsRNA Add dsRNA of another gene of interest (eg.CG4091, rpr) 1.Gene knock-out of iap1(anti-cell death gene) causes apoptosis 2.Knock-out of genes upstream of iap1in the same pathway and the genes that do not interact withiap1will still induce apoptosis Apoptotic body Dead cell Cell forming apoptotic body Tryphan blue stain 1.Knock-out of genes down- stream ofiap1in the same pathway will inhibitiap-1 induced apoptosis “Double knock-out” strategy Mbn2 cell line Addiap1 dsRNA Add dsRNA of another gene of interest (eg.CG4091, rpr) 1.Gene knock-out of iap1(anti-cell death gene) causes apoptosis 2.Knock-out of genes upstream of iap1in the same pathway and the genes that do not interact withiap1will still induce programmed cell death Apoptotic body Dead cell Cell forming apoptotic body Tryphan blue stain 1.Knock-out of genes down- stream ofiap1in the same pathway will inhibitiap-1 induced programmed cell death. A “Double knock-out” strategy Mbn2 cell line Addiap1 dsRNA Add dsRNA of another gene of interest (eg.CG4091, rpr) 1.Gene knock-out of iap1(anti-cell death gene) causes apoptosis 2.Knock-out of genes upstream of iap1in the same pathway and the genes that do not interact withiap1will still induce apoptosis Apoptotic body Dead cell Cell forming apoptotic body Tryphan blue stain 1.Knock-out of genes down- stream ofiap1in the same pathway will inhibitiap-1 induced apoptosis “Double knock-out” strategy Mbn2 cell line Addiap1 dsRNA Add dsRNA of another gene of interest (eg.CG4091, rpr) 1.Gene knock-out of iap1(anti-cell death gene) causes apoptosis 2.Knock-out of genes upstream of iap1in the same pathway and the genes that do not interact withiap1will still induce programmed cell death Apoptotic body Dead cell Cell forming apoptotic body Tryphan blue stain 1.Knock-out of genes down- stream ofiap1in the same pathway will inhibitiap-1 induced programmed cell death.

Drosophila l(2)mbn cell line established in 1978 by Gateff consists of tumorous haemocytes isolated from a larva of the Drosophila mutant lethal (2) malignant blood neoplasm. form vacuoles and die in response to 20-hydroxyecdysone (20HE; ecdysone) treatment die in response to treatment with Diap1-RNAi morphology and gene expression changes currently under investigation

RNAi screen design Prepare dsRNA using T7-tailed gene specific primers (average product size = 500 bp) Add approx 50nM dsRNA directly to Drosophila l(2)mbn cells under serum-free conditions & incubate 1 hr. Add serum. Incubate 4-5 days Cell counts/WST-1 colorimetric assay (cell viability) Microscopic observation (cell morphology) No treatmentEcdysone treatment Diap1-RNAi treatment

Concept Induction of death Ecdysone Diap1-RNAi Required genes RNAi

PCD pathways in Drosophila (Meier et al., Nature 2000)

CG4091-RNAi partially blocks PCD induced by Diap1-RNAi RNAi construct No. of live cells

PCD pathways in Drosophila (Meier et al., Nature 2000) CG4091 TNF?

CG4091: work in progress P-element excision (flybase.bio.indiana.edu) Overexpression and loss-of-function in vivo siRNA in mammalian cells RNAi in Drosophila l(2)mbn cells

Summary 72hr Untreated Control Cells 72hr 10µM 20-hydroxyecdysone Treated Cells 72hr Untreated Control Cells 72hr 10µM 20-hydroxyecdysone Treated Cells 1244 / 4628 transcripts differentially expressed prior to PCD – stage and tissue specificity of starting material represents a highly enriched source for detection of gene expression differences. Many have candidate human orthologs differentially expressed in cancer. There is overlap between apoptosis and autophagic PCD with respect to the genes involved (e.g. known apoptosis genes detected in our differentially expressed genes; a few of these were known previously but we detected others not previously described in the salivary gland). There appear to be genes specific to autophagic PCD – eg. putative autophagy gene orthologs and lysosomal genes were differentially expressed prior to autophagic PCD. An RNAi system for functional characterization of candidates has been designed.

Comparison of SAGE and real-time quantitative RT-PCR Fold-difference by SAGE Fold-difference by QRT-PCR Correlation coefficient = 0.5 II. Correlation coefficient between fold-difference values (64 samples): I. Direction of Change: 91/96 samples = 95% concordance

Tag-to-gene Mapping in Drosophila (E. Pleasance, M. Marra and S. Jones, submitted) AAAAA CATGAGGAGTGAAT Gene X Platform: Queryable ACEDB database Resources: Drosophila genomic sequence and annotation (GadFly Release 2) predicted UTRs 259,620 ESTs and full-length cDNAs (BDGP) 5,181 salivary gland 3’ ESTs (GSC)

Programmed cell death Type I= Apoptotic cytoskeletal collapse condensation and fragmentation of chromatin and cytoplasm preservation of organelles phagocytosis by macrophages or neighbouring cells (heterophagy) occurs in isolated cells Type II = Autophagic preservation of cytoskeleton formation of vacuoles that engulf cytoplasm and organelles fusion of vacuoles with lysosomes for self-degradation late chromatin condensation and nuclear degeneration occurs in groups of cells

Mining Expression Data in Drosophila I. Keyword-based Data Mining e.g. Keyword cancer: 33 associations GadFly – Swissprot Homology Table Extract entries based on keyword search II. Cross-species Gene Expression Comparisons CG4091 upregulated 102-fold in 16 vs 23 hr salivary glands SCC-S2 downregulated 7-fold in human mammary gland ductal carcinoma vs normal Drosophila autophagic cell death Human cancer

Mining Expression Data in Drosophila I. Keyword-based Data Mining Q: How many differentially expressed Drosophila genes have a mammalian homolog that has been associated with cancer? A: 33 genes GadFly – Swissprot Homology Table Extract entries based on keyword search

Keyword Data base SG 16 SG 23 Gene Sim (%) Score Length (aa) Swiss Prot id Database Description death apoptosis death survival autophagy hormone cancer tumor apoptosis TNF BH3 SP FB PCD EST only Nc debcl cact Traf1 CG4719 Ptpmeg CG4859 CG10777 Rpn2 stck chrw ciboulot CG8706 Atet CG13907 botv sp6 CG11335 CG10990 CG4091 CG ICE6_HUMAN BCL2_HUMAN IKBA_HUMAN TRA1_HUMAN BAR1_HUMAN PTND_HUMAN MM11_MOUSE WRN_HUMAN PSD2_HUMAN PINC_HUMAN RB24_MOUSE TYB4_HUMAN LRP2_HUMAN ABG2_HUMAN MOT1_HUMAN EXL3_HUMAN MASP_MOUSE LYOX_HUMAN CASPASE-6 PRECURSOR APOPTOSIS REGULATOR BCL-2 NF-KAPPAB INHIBITOR ALPHA TRAF1 BRCA1-ASSOCIATED RING DOMAIN FAS-ASSOCIATED PTP-1 STROMELYSIN-3 PRECURSOR WERNER SYNDROME HELICASE 26S PROTEASOME S2 PINCH PROTEIN RAS-RELATED PROTEIN RAB-24 THYMOSIN BETA-4 LDL RECEPTOR-RELATED PROT BREAST CANCER RESISTANCE MONOCARBOXYLATE TRANSPORT TUMOUR SUPPRESSOR EXL3-LIKE PROTEASE INHIBITOR 5 PROT-LYSINE 6-OXIDASE PREC. MM 'APOPTOSIS PROTEIN MA-3' HS 'TNF-INDUCED PROTEIN GG2-1' BH3

Mining Expression Data in Drosophila II. Cross-species Gene Expression Comparisons CG4091 upregulated 102-fold in 16 vs 23 hr salivary glands SCC-S2 downregulated 7-fold in human mammary gland ductal carcinoma vs normal Drosophila autophagic cell death Human cancer (CGAP)

What are l(2)mbn cells? Tumorous haemocyte derived from a Drosophila mutant lethal malignant blood neoplasm. Established in 1978, isolated from mutant larvae Non-homogenous mixture of cells: 3 types 1) plasmatocyes, 2) lamellocytes, 3) podocytes

RNAi screen results Genes tested with RNAi: CG40228 CG11051 Sgs7 Talin CG8785 CG14214 Diap1 CG9911 CG7220 Timp CG7059 Sema-5c CG4798 Possible phenotype: CG14214 Cell Death Phenotype: diap1

WST-1 Assay Colourmetric assay, absorption at 450nm Viable cells with mitochondrial dehydrogenase activity cleave a Water-Soluble Tetrazolium salt (WST-1) to a dark red water-soluble formazan Conversion of WST-1 to formazan correlates directly with the number of metabolically active cells

WST-1 Assay Use to determine the amount of live cells present in non-treated control cells and 20HE treated cells Expect a difference in absorption readings between 20HE treated and non-treated cells Fast, easy screening assay for RNAi screen

WST-1 Assay Results Decrease in cell respiration of 20HE treated cells – Directly represents the amount of viable cells

Programmed cell death Apoptosis –Caspase-dependent, chromatin and cytoplasm condensation –Triggered by DNA damage, oncogene activation, extracellular signals, etc. Autophagic cell death –Morphologically similar to autophagy (triggered by nutrient starvation) –Vacuoles formed containing cellular components –Triggered by Ras activation, ??

Finding PCD genes by orthology and expression Human cancer SAGE libraries Differentially expressed genes Drosophila orthologues Drosophila SAGE library Differentially expressed genes Human orthologues Set of Drosophila/human orthologues perturbed in both cancer and Drosophila PCD

Differentially expressed gene sets Genes differentially expressed between 16h and 23h in Drosophila salivary gland 361 genes upregulated 203 genes downregulated Genes differentially expressed between normal and cancer SAGE libraries from CGAP (Unigene) 2277 genes Find human RefSeq orthologs 197/361 upregulated 99/203 downregulated Map to LocusLink, Unigene Find common genes: Downregulated in PCD: 7 Drosophila genes, 8 human orthologs Upregulated in PCD: 16 Drosophila genes, 18 human orthologs

Genes downregulated in PCD

Autophagic PCD and cancer Autophagic PCD is associated with the following: spontaneous regression of human neuroblastoma (Kitanaka et al. 2002) tamoxifen-treated human breast cancer cells, MCF-7 (Bursch et al. 1996) tamoxifen-treated murine breast cancer cells, FM3A (Bilir et al. 2001) TNFa-treated T lymphoblastic leukaemic cells (Jia et al. 1997) bcl-2 antisense treated human leukemic HL60 cells (Saeki et al. 2000) oncogenic Ras-expressing human glioma and gastric cancer cells (Chi et al. 1999)

PCD and Cancer The role of apoptosis in cancer is well established: “resistance toward apoptosis is a hallmark of most and perhaps all types of cancer” (Hanahan & Weinberg, Cell, 2000) Autophagic PCD and autophagy have been associated with cancer recently:

CG4091-RNAi partially blocks PCD induced by Diap1-RNAi

Autophagy and cancer Beclin-1 (apg6) is an autophagy gene: Monoallelically deleted and expressed at reduced levels in human breast and ovarian cancers beclin1 knockout mouse indicated that beclin1 is a haploinsufficient tumor suppressor gene; heterozygotes displayed an increase in the incidence of lymphoma, lung carcinoma and liver carcinoma (Qu et al. 2003; Yue et al. 2003) Current notion: “Defective autophagy can lead to cancer” (Edinger & Thompson, 2003)

PCD and disease Function Dysfunction Deleting damaged cells Cancer Culling cell number Deleting structures Sculpting tissues Autoimmune diseases Neurodegenerative diseases Developmental abnormalities

Research Aims 1.Identify the genes involved in autophagic cell death in vivo. 2.Determine which genes are necessary and sufficient for autophagic cell death. 3.Identify the autophagic cell death genes/pathways associated with human disease and investigate potential as molecular markers and/or therapeutic agents.

siRNA analysis of autophagic cell death genes in human breast carcinoma cells Overview of experimental design Test for RNA depletion by RT-PCR Preparation and storage of siRNAs corresponding to selected human candidate autophagic cell death genes MCF-7 cells + Reporter +/- Tamoxifen MicroscopyCell Death AssaysAutophagic vacuole assays 3-4 days incubation time to ensure protein depletion Test candidate genes for expression in MCF-7 by RT-PCR Transfection (siPORT Lipid, Ambion)

Specific Aim Identify and characterize genes that are necessary for mammalian autophagic cell death. Experimental Approach RNAi analysis of candidate autophagic cell death genes in human breast carcinoma cell line MCF-7.

Autophagic cell death in disease human neurodegenerative diseases (Alzheimer and Parkinson) cardiomyocyte degeneration spontaneous regression of human neuroblastoma tamoxifen-treated mammary carcinoma cells (MCF-7) bcl-2 antisense treatment of human leukemic HL60 cells beclin-1 (apg6) promotes autophagy and inhibits tumorigenesis; expressed at decreased levels in human breast carcinoma

GSC Programmed Cell Death Group Ian BosdetAmy Leung Suganthi ChittaranjanMelissa McConechy Doug FreemanErin Pleasance Claire HouMohammed Qadir Jennifer Kouwenberg Marco Marra Victor Ling Sharon Gorski

Drosophila expression analysis data SAGE libraries for PCD gene identification from salivary gland (Sharon Gorski, Suganthi Chittaranjan, Doug Freeman) 3 SAGE libraries (16h, 20h, 23h) hr (APF, 18°C) RT diap2 rpr hid (S. Gorski et al. Curr. Biol. 13: , 2003)

Determining orthology Find orthologs using InParanoid (Remm et al, 2001) –Based on BLAST similarity –Groups genes arising from duplications – eg. one Drosophila gene may be orthologous to multiple human genes Find orthologs between Drosophila genes (from GadFly/FlyBase) and human RefSeq sequences