1. Linear motifs and phosphorylation sites

2. What is a linear motif? ( in molecular biology )

3. Short sequence of amino acids encoding a particular molecular function …a first taste We need a more accurate definition! Linear Motifs Functional sites

4. What are you going to learn about Linear Motifs? Why are they important? Where can we find them? How can we discover them? When and how can we use them? What are tools and resources to handle them? Can we classify them? How can we represent them?

5. What are you going to learn about Linear Motifs? Why are they important? Where can we find them? How can we discover them? When and how can we use them? What are tools and resources to handle them? Can we classify them? How can we represent them?

6. Tyrosine kinsase Src has several functional sites CSK phosphorylation (Y527) & SH2 ligand SH3 ligand Auto phosphorylation site (Y416) Myristoylation site

7. MDM2 TAFII31 P300 NLS CYCLIN CBP NES S100B SIR2 phosphorylation Pin1 P-Ser-Pro isomerisation Acetylation SUMO Ubiquitinylation p53 is full of functional sites

8. The sequences of many proteins contain short, conserved motifs that are involved in recognition and targeting activities, often separate from other functional properties of the molecule in which they occur. Tim Hunt (TIBS 1990) These motifs are linear, in the sense that three-dimensional organization is not required to bring distant segments of the molecule together to make the recognizable unit.

9. Tim Hunt (TIBS 1990) The conservation of these motifs varies: some are highly conserved while others, for example, allow substitutions that retain only a certain pattern of charge across the motif.

10. A more accurate definition short, common stretches of polypeptide chains (~ 3-10 amino acid residues long) embody a distinct molecular function independent of a larger sequence/structure context. are nearly always involved in regulation are involved in protein/domain-protein/domain interactions often reside in disordered or low-complexity regions often become ordered upon binding to another protein or domain bind with low affinity (1.0-150  M). Mediate transient interactions. occurrences of LMs seem to arise or disappear as a result of point mutations

11. What are you going to learn about Linear Motifs? Why are they important? Where can we find them? How can we discover them? When and how can we use them? What are tools and resources to handle them? Can we classify them? How can we represent them?

12. Evolutionary unrelated protein sharing a functional feature are likely to contain similar linear motifs This may be the result of - convergent evolution - evolutionary conservation in a divergent evolution process Why are they important? In any case, linear motifs are indicative of functions With the appropriate tools, they can be used to identify: protein functions functional regions (in a protein sequence and on its three- dimensional structure, if available) They are made up of the amino acid residues encoding a functional site In other words…

13. What are you going to learn about Linear Motifs? Why are they important? Where can we find them? How can we discover them? When and how can we use them? What are tools and resources to handle them? Can we classify them? How can we represent them?

14. Can we classify LMs? How?

15. Functional group Functional site (Linear Motif)

16. PRACTICE: Let’s find linear motifs in human p53… Go to the UniProt website: http://www.uniprot.org/ Type p53 in the Query text box and select P04637 or Type directly either P04637 or P53_HUMAN in the Query text box Work in groups and analyse the p53 entry record: - how many LMs can you identify? - which function(s) are they indicative of? - are they always annotated as “motif”? - can you classify them according to the 4 categories?

17. What are you going to learn about Linear Motifs? Why are they important? Where can we find them? How can we discover them? When and how can we use them? What are tools and resources to handle them? Can we classify them? How can we represent them?

18. How can we represent LMs? Regular expression: [RK].L.{0,1}[FLIV] inhibitors Alignment of cyclin ligands

19. How can we represent LMs? Regular expression: [RK].L.{0,1}[FLIV] inhibitors Alignment of cyclin ligands

20. Regular Expression (regexp) L: single amino acid “L” = Leucine [KR]: different amino acids allowed at this position x or.: wildcard {0,1}: variable length

21. Regular Expression: Examples

22. Before we describe what regexp are useful for, let’s briefly see how to discover de novo motifs In some cases, the structure and function of an unknown protein which is too distantly related to any protein of known structure to detect its affinity by overall sequence alignment may be identified by its possession of a particular cluster of residues types classified as a motifs. The motifs, or templates, or fingerprints, arise because of particular requirements of binding sites that impose very tight constraint on the evolution of portions of a protein sequence Arthur Lesk, 1988

23. What are you going to learn about Linear Motifs? Why are they important? Where can we find them? How can we discover them? When and how can we use them? What are tools and resources to handle them? Can we classify them? How can we represent them?

24. In contrast to domains, which are readily detectable by sequence comparison, linear motifs are difficult to discover due to their short length, a tendency to reside in disordered regions in proteins, and limited conservation outside of closely related species. Neduva et al. PLoS Biology 2005

25.  Study literature paper(s)/review(s) on a group of unrelated proteins sharing a function  Build an alignment of these proteins  Add to the alignment other sequences relevant to the subject under consideration  Pay attention to the residues and regions thought or proved to be important to the biological function of that group of proteins: enzyme catalytic sites PTM sites regions involved in binding  Try to find a short conserved sequence which includes functionally important residues De novo Linear Motif discovery

26. Discovery of de novo Linear Motif There are algorithms that do it automatically Neduva et al. PLoS Biology 2005

27. Discovery of de novo Linear Motif Neduva et al. PLoS Biology 2005 Our central hypothesis is that proteins with a common interaction partner will share a feature that mediates binding, either a domain or a linear motif. In the absence of a shared domain, a linear motif could well be the only common sequence feature and might thus be detectable simply by virtue of over-representation, which is the basis of our approach.

28. Edwards et al. PLoS ONE 2007 A probabilistic method for identifying over-represented, convergently evolved, short linear motifs in proteins.

29. PRACTICE: Discovery of de novo Linear Motifs http://dilimot.russelllab.org/ http://www.southampton.ac.uk/~re1u06/software/slimfinder/ Dilimot SLIMFinder

30. What are you going to learn about Linear Motifs? Why are they important? Where can we find them? How can we discover them? When and how can we use them? What are tools and resources to handle them? Can we classify them? How can we represent them?

31. Linear Motif Databases PROSITE ELM 1632 documentation entries (domains and functional sites) 174 manually annotated motifs 16-03-2012 R-x-[RK]-x(1,2)-R R.[RK]{1,2}.R

32. How can we use regular expressions? Regular expressions can be used to search for motif occurrences in (uncharacterised) protein sequences There are algorithms that do this for us A motif (a regexp) can have many instances We call the occurrence of a motif in a sequence an INSTANCE of that motif What regular expressions are useful for? KKVAVVRTPPKSPSSAKSRL ISPPTPKPRPPRPLPVAPGS EDQILKKPLPPEPAAAPVST SHRKTKKPLPPTPEEDQILK TRICKIYDSPCLPEAEAMFA [RKY]..P..P TAU_HUMAN P85A_HUMANBTK_HUMANBTK_HUMAN RAD51_HUMAN SH3 ligand motif

33. Prediction of new instances of Linear Motifs ScanProsite Scansite ELM MiniMotifMiner Allows the search for user-defined regular expressions INPUT: a protein sequence OUTPUT: PROSITE or user-defined motif matches in the input sequence INPUT: a protein sequence OUTPUT: scansite motif matches in the input sequence INPUT: a protein sequence OUTPUT: ELM motif matches in the input sequence INPUT: a protein sequence OUTPUT: MiniMotifMiner motif matches in the input sequence

34. PRACTICE: Prediction of new instances of Linear Motifs http://prosite.expasy.org/scanprosite/ Go to the ScanProsite website and search for the RGD motif in the SwissProt database How many hits? How many hits are expected by chance? R-G-D Select database

35. Regular expression pros and cons AdvantagesDisadvantages Memorable to humansOver determined Computationally fastMotif may vary in other lineages Standardised in scripting languages (Python, Perl) Do not capture weaker preferences Often, they can descrive a motif very well Easy to make a poor representation Unfortunately matches to these motifs are not significant, providing a signal-to-noise problem for bioinformatics tools

36. Overprediction and context information

37. Functional sites only work in proper context The cell knows how to discriminate TP from FP !!! The site must be in the correct cellular context cellular context (subcellular localisation) (subcellular localisation) The site is only relevant in a specific taxonomy range taxonomy range Knowledge of context can provide the basis for filters for improved prediction of functional sites The site must be in correct molecular context context - accessible - accessible - usually not in globular domains, - usually not in globular domains, - often together with certain types of co-domains - often together with certain types of co-domains

38. For example…

39. Motifs are mostly found in disordered regions Globular domain filter Src kinase The disordered regions are proving to be rich in Linear Motifs We can exploit this observation and filter out motif matches inside domains

40. When inside a domain, a motif match is more likely to be a True Positive (TP) if it occurs in a flexible (i.e. loop, turn or linker) and accessible region of the domain Structural Filter Inside domains they are unlikely unless in surface loops Motif matches are not ALWAYS outside domains

41. An exposed instance of the RGD motif in a domain An instance of the RGD motif in a region outside a domain The RGD motif is recognized by different members of the integrin family

42. Two MOD_N-GLC_1 motifs in a domain MOD_N-GLC_1 (.(N)[^P][ST]..) is a motif for N-glycosilation site

43. We can think to implement a filter that is based on the three-dimensional features of motifs (i.e. their accessibility and secondary structure types) If the match is not accessible If the match is in  -helix If the match is in  -strand low score Structural Filter

44. Other features that can be used to filter out FPs: Taxonomy Cellular compartment Evolutionary conservation Davey NE et al. Mol Biosyst 2011

45. Improve the prediction of LM instances by discarding those matches that are unlikely to be functional because they have not been conserved during the evolution of the protein sequences Why is a Conservation Score useful for linear motif prediction?

46. There is a resource which implements these filters It associates a score to occurrences of motifs based on Cellular context Molecular context Domain context Disorder Taxonomy Evolutionary conservation

47. The Eukaryotic Linear Motif (ELM) Resource implements a logical filtering system to reduce false matches

48. The Eukaryotic Linear Motif (ELM) Resource Repository of information about functional sites (including experimentally reported instances) A motif-based query tool to find possible new functional sites A logical filtering system to reduce false matches

49. The ELM Resource - An overview

50. PRACTICE: The ELM server (http://elm.eu.org/) Go to the ELM server Search for motif matches in the EH domain-binding mitotic phosphoprotein

51. Output 1 annotated instance Instance in unfavourable context instance in structurally unfavourable context highly conserved instance

52. Output 2

54. Browse the ELMs page for the Clathrin Box motif in Endocytosis cargo adaptor proteins (ELM: LIG_AP2alpha_2)

55. Link to reported instances

58. Exploring unknown protein sequences

60. Phosphorylation sites

61. Phosphorylation is the addition of a phosphate group (PO 4 ) to a protein molecule or small molecule. The hydroxyl groups (-OH) of SER, THR or TYR residues side chain are the most common targets

62. A protein kinase moves a phosphate group from ATP to the protein A protein phosphatase removes the phosphate and the protein reverts to its original state. ATP (adenosine triphosphate) is the energy currency of the living world. Every cellular process that requires energy gets it from ATP It is rapid (few seconds) It is easily reversible Reversible protein phosphorylation

63. It is involved in regulation of metabolism, motility, growth, division, differentiation, trafficking, membrane transport, learning, memory ~ one third of cellular proteins could undergo phosphorylation Even subtle changes in the activity of protein kinases can lead to a variety of diseases (cancer) Reversible protein phosphorylation regulates most aspects of cell life

64. Phosphorylation is a Post Translational Modification (PTM) A kinase recognises its substrate and adds a phosphate group (PO 4 ) to one of its residues, typically a Serine (Ser, S), Threonine (Thr, T), or Tyrosine (Tyr, Y) Amino acid phosphorylation is probably the most abundant of the intracellular PTMs used to regulate the state of eukaryotic cells, with estimates ranging up to 500,000 phosphorylation sites in the human proteome

65. Substrate recognition is specific Each kinase is capable of recognising its substrate(s) in the cell In other words… Nevertheless… Even though the determinants of specificity are still unclear In fact, the enzymes must be specific and act only on a defined subset of cellular targets to ensure signal fidelity.

66. Substrate recruitment is one of the known specificity mechanisms The protein composition around the phosphorylatable site is another factor Kinases are capable of recognising the region surrounding the phosphoacceptor residue (in sequence and/or in structure) In fact, kinases do not phosphorylate every Ser, Thr, Tyr they encounter in the cell Kreegipuu et al, NAR 1998

67. A phosphorylation site can be represented by a phosphorylation motif Experimentally verified phosphorylation motifs can be used to predict new phosphorylation sites and characterise kinase substrates

68. There are many resources collecting P-sites and many tools to predict P-sites in user-defined protein sequences Collection of instances of P-sitesPrediction of new instances of P-sites Phospho.ELM phospho.elm.eu.org/ Phospho.ELM phospho.elm.eu.org/ PhosphoSitePlus www.phosphositePlus.org/ Scansite scansite.mit.edu/ PHOSIDA www.phosida.com/ NetPhos www.cbs.dtu.dk/services/NetPhos/ PHOSPHORYLATION SITE DATABASE www.phosphorylation.biochem.vt.edu/ NetPhosK www.cbs.dtu.dk/services/NetPhos/ Phospho.3D www.phospho3d.org/ NetworKIN networkin.info/search.php KinasePhos KinasePhos.mbc.nctu.edu.tw/ Predikin predikin.biosci.uq.edu.au/

69. Current release contains: 42,914 instances (fully linked to literature references) 299 kinases 11,224 sequences 8,698 substrates Phospho.ELM phospho.elm.eu.org Database of experimentally verified phosphorylation sites in eukaryotic proteins

70. PRACTICE Go to the Phospho.ELM website and search P-sites for p53

73. ELM and Phospho.ELM are interconnected

74. PhosphoBlast

76. Structural information on P-sites and 3D scan

77. Phospho.3D http://www.phospho3d.org/ PRACTICE Go to the Phospho.3D website and search all the substrates of the Src kinase

81. MEESQSDISLELPLSQETFSGLWKLLPPEDILP SPHCMDDLLLPQDVEEFFEGPSEALRVSGAPA AQDPVTETPGPVAPAPATPWPLSSFVPSQKTY QGNYGFHLGFLQSGTAKSVMCTYSPPLNKLF CQLAKTCPVQLWVSATPPAGSRVRAMAIYKKS QHMTEVVRRCPHHERCSDGDGLAPPQHLIRV EGNLYPEYLEDRQTFRHSVVVPYEPPEAGSEY TTIHYKYMCNSSCMGGMNRRPILTIITLEDSSG NLLGRDSFEVRVCACPGRDRRTEEENFRKKE VLCPELPPGSAKRALPTCTSASPPQKKKPLDG EYFTLKIRGRKRFEMFRELNEALELKDAHATEE SGDSRAHSSYLKTKKGQSTSRHKKTMVKKVG PDSD Suggestions to predict P-sites in unknown sequences ?

82. Go to UniProt (or Blast your sequence against the UniProt database) and explore the sequence annotation Go to Phospho.ELM and scan the sequence Go to PHOSIDA and PhosphoSitePlus and do the same Use different predictors and select only high scoring sites Use structural information if available: - is the site exposed? - is it in a flexible region? Use domain (SMART and Pfam) databases: - is the site inside a domain? Use evolutionary information: - is the site conserved? Exploring unknown protein sequences

83. When all information is collected, only retain sites predicted by more than one tool Not inside domain(s) Not in secondary structure elements (helices and strands) Accessible to the solvent Evolutionary conserved Amongst these, for further experimental tests, preferably choose sites that are: Exploring unknown protein sequences