1. bsrweb.burnham.org Stacy Xiayu Huang Roy Williams Alexey Eroshkin Gene-centric bioinformatics analysis to guide your cancer research

2. bsrweb.burnham.org Bioinformatics Shared Resource (BSR)/ Information and data management (IDM) Stacy Xiayu Huang, Biostatistician Roy Williams, Bioinformatics Specialist Alexey Eroshkin, Bioinformatics Specialist and Director

3. Guess what gene will we explore? bsrweb.burnham.org

4. CASP8 Caspase-8 bsrweb.burnham.org

5. Caspase-8 Cysteine proteases that plays a critical role in the regulation of programmed cell death (apoptosis). Present in the cytosol as inactive zymogen that become activated in response to specific death stimuli. Once activated, initiator caspases (caspase-8, -9, and -10) cleave and activate executioner caspases (caspases-3 and -7). Activity post-translationally regulated Two primary pathways to establish the cell death program: intrinsic mediates response to cellular stress, such as DNA damage, and extrinsic, triggered by extracellular signals such as Fas binding to its cognate receptor and leads to activation of initiator caspase-8. Highly conserved active sites and overlapping substrate specificities make it difficult to use inhibitors or activity-based probes to study the function, activation, localization and regulation of individual members of this family. bsrweb.burnham.org

6. Today: Assessing the potential impact of variations. Mutations in caspase-8 substrates. Cancer-related driver mutations vs passenger mutations Changes in a gene's expression during tumorigenesis. Including microarray, RNA-seq, chIP-seq and methylation data. Mutations and cancer stages, drug treatments and developing drug resistance * Most tools from web – public and proprietary Sometimes program install required We are NOT doing a review of casp8 publications We are NOT providing an expert opinion on casp8

7. Caspase-8 in pdb – very graphical bsrweb.burnham.org Death effector domains (DEDs) are protein-protein interaction structures that are found in proteins that regulate a variety of signal transduction pathways.

8. MetaCore - for every gene: bsrweb.burnham.org

9. Casp8 in Metacore (Thomson Reuters): pathways and processes bsrweb.burnham.org

10. MetaCore: Apoptosis and survival Caspase cascade bsrweb.burnham.org

11. MetaCore’s Process Network: Apoptosis_Apoptosis stimulation by external signals Interactions Hubs Divergence hubs Convergence hubs Nodes Transcription factors Membrane receptors Secreted proteins & peptides GO processes Diseases Tissues

12. You can edit Process Networks Change layout Add object Remove/move object Add relation/connection from your experiment Get more info for every –Relation –Object bsrweb.burnham.org

13. MetaCore: associated diseases (part)

14. Casp8: number of publications (associated diseases) 60 other associated diseases have just one reference: e.g., Adenocarcinoma, Bronchiolo-AlveolarAdenomaAdenomatous Polyposis ColiAIDS-Related ComplexAnoxia Arthritis, RheumatoidCarcinomaCarcinoma, Basal CellCarcinoma, Ductal, BreastCarcinoma, Lobular Carcinoma, Small CellClubfootDermatitis, Allergic ContactEndometrial Neoplasms Ependymoma Esophageal NeoplasmsGallbladder Neoplasms, etc

15. Drugs and biomarkers bsrweb.burnham.org

16. Summary on CASP8 in UniProt (well respected database) (http://www.uniprot.org/uniprot/Q14790) (http://www.uniprot.org/uniprot/Q14790

17. Sequence annotation in UniProt bsrweb.burnham.org

18. Switch between natural variants and isoforms can be related to cancer (UniProt) L (long): missing the catalytic site of caspase-8 but retains 2 N- terminal repeats of the death-effector domain – inhibitor of casp8

19. You can align isoforms sequences at UniProt and find that isoforms 5,6,7 and 8 lack catalytic sites.

20. bsrweb.burnham.org CASP8 in ENTREZ

21. CASP8 in Genome Browser from NextBio Not showing protein changes for SNPs but can show your study in context of others

22. bsrweb.burnham.org Protease association with cancer* Gene variants and isoforms Variation effects –in caspase-8 Inactivation, partial or complete altered signaling (different phosphorylation) Change in stability Change of specificity (e.g., new substrates) Change in Interactions localization –in caspase-8 targets Known targets may be not cleaved (not activated) or cleaved in a wrong place –new substrates can appear (cleavage sites in other proteins) Inactivating some protein functions Creating new activity Deregulation (epigenetics or transcriptional or posttranslational) –Up regulation - increase apoptosis, can cleave beyond targets –Down regulation - inhibits apoptosis (cells not dying … cancer?)

23. bsrweb.burnham.org How to test if there is a connection between CASP8 isoforms and cancer Do Next-gen sequencing (RNA-seq) Calculate transcription-level expression of gene’s isoforms (Partek, E/M) Expectation maximization (E/M) algorithm is used to quantitate gene alternative splicing in two sets of samples Assigns significance of the change in isoforms expression Here is the power of Next-Gen sequencing Cufflinks or SLIDE algorithms will do as well

24. bsrweb.burnham.org Mutations: Check first if there are somatic cancer- related mutations in your gene - List of ~500 genes associated with Cancers (Cancer Somatic Mutations) - Casp8 not in the “Catalog”

25. The catalog of cancer genes (part) bsrweb.burnham.org

26. Cosmic: 22 somatic mutation in casp8 bsrweb.burnham.org

27. But: very small % of tissues with mutated casp8 bsrweb.burnham.org

28. Known clinical variants at NCBI : http://www.ncbi.nlm.nih.gov/clinvar/?term=casp8 http://www.ncbi.nlm.nih.gov/clinvar/?term=casp8

29. bsrweb.burnham.org Ways to assess the significance of the amino acid variation on protein function Most algorithms assume that important positions in a protein sequence have been conserved throughout evolution and therefore substitutions at these positions may affect protein function. SIFT, popular non-synonymous SNP annotation software that assigns a "functional importance" score to SNPs, http://sift.jcvi.org/http://sift.jcvi.org/ PolyPhen, PolyPhen2: http://genetics.bwh.harvard.edu/pph2/http://genetics.bwh.harvard.edu/pph2/ DIY: Align protein sequences and analyze the replacement on your own – use your best judgment, look for 3D structure –Use family sequence alignment –Assess variability of residues in particular position –Role of the position in structure/function –If mutation outside of known in mammalian species, likely destructive

30. Where to go: Merops and CutDB – source of data on proteases, protease known substrates

31. Assess mutation in coding region using CASP8 multiple sequence alignment http://merops.sanger.ac.uk/cgi-bin/code_align?mid=C14.009

32. Cut and paste alignment into JalView and you can see the natural variations JalView can be installed or webstart can be used

33. bsrweb.burnham.org Can dig deep – to 3D structure

34. bsrweb.burnham.org To analyze ALL KNOWN mutations go to Exome Sequencing Project (NHLBI ESP) http://evs.gs.washington.edu/EVS/

35. bsrweb.burnham.org dbSNP database http://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?rs=17860424 http://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?rs=17860424

36. bsrweb.burnham.org How to evaluate the effect of mutations in CASP8 substrates (cleavage sites in cancer sample) Motif: S4 S3 S2 S1 S1’ [DL][ES] X D* [GSA] Position weight matrix

37. Assess variations in cancer genome for new casp8 substrates Predict casp8 cleavage sites in cancer sample proteins (e.g., http://www.dmbr.ugent.be/prx/bioit2- public/SitePrediction/, http://www.proteolysis.org/)http://www.dmbr.ugent.be/prx/bioit2- public/SitePrediction/http://www.proteolysis.org/ Predict casp8 cleavage sites in normal sample proteome Compare two predictions and analyze New cleavage sites in proteins from cancer samples Missing cleavage sites in proteins from cancer samples Prioritize cancer relevance of the difference (COSMIC gene set, cancer-related pathways etc.)

38. PMAP whole proteome substrate prediction (part) bsrweb.burnham.org

39. Casp8 cleaves phosphorylated sites and cleavage sites get phosphorylated caspase cleavage can expose new sites for phosphorylation, and, conversely, Phosphorylation at the +3 position of cleavage sites can directly promote substrate proteolysis by caspase-8 functional crosstalk between phosphorylation and caspase proteolytic pathways that lead to enhanced rates of protein cleavage and the unveiling of new sites for phosphorylation phosphorylation events are enriched on cleaved proteins and are clustered around sites of caspase cleavage Interesting to compare cleavage in phosphorylated and non phosphorylated proteins Melissa M Dix, Cell 150, 426–440, July 20, 2012 ª201

40. Human phospho-site knowledge database: casp8 (http://www.phosphonet.ca) bsrweb.burnham.org

41. You study some cancer and you want to find cancer driver genes Perform whole-exome sequencing –matched normal DNAs and –metastatic tumor DNAs Variant detection (next-gen sequencing) and validation (Sanger) Find genes that appeared to be somatically mutated at elevated frequency (statistical model) –Majority (many) of tumor samples have disruptive variations –Few or no normal sample have such variations bsrweb.burnham.org Wei X. Nat Genet. 2011 May;43(5):442-6

42. Identification of likely driver mutations –Take all (or frequent) somatic mutations –Filter for the cancer genes from the Cancer Gene Census (http://www.sanger.ac.uk/genetics/CGP/Census/) that are known to be mutated to contribute to cancer development –Take mutations that conformed to the known patterns of cancer- causing mutation for each cancer gene in COSMIC database Truncating mutations, missense mutations, essential splice site mutations and homozygous deletions for recessive cancer genes For dominant genes, include mutations that had been previously registered in COSMIC –CONS: this cover only 2% of human genes bsrweb.burnham.org Stephens, Nature, 486, p.400, 2012

43. Identify frequent somatic drivers Frequency of mutation (cancer vs normal samples) –Tumor tissue with mutation (T) –Normal and Tumor tissues with mutation (N+T) –T/(N+T) ratio –Consider T/(N+T) > cutoff (0.2?) –For this mutation find MAF - minor allele frequency in population (ESP6500 data) MAF 0.01 considered as common polymorphism) –Compare T/(N+T) ratio with MAF T/(N+T) >> MAF – potential driver mutation Example: T/(N+T) =0.2, MAF = 0.005 bsrweb.burnham.org

44. CASP8: Driver mutations implicated in breast cancer development bsrweb.burnham.org Stephens, Nature, 486, p.400, 2012

45. The Human Protein Atlas http://www.proteinatlas.org/ protein expression profiles based on immunohistochemistry for a large number of human tissues, cancers and cell lines bsrweb.burnham.org

46. Casp8 antibody staining in breast cancer bsrweb.burnham.org

47. Thanks Questions? bsrweb.burnham.org