PART II.PART II.Prediction of functional Prediction of functional regions within disordered regions within disordered proteinsproteins

Zsuzsanna DosztányiMTA-ELTE Momentum Bioinformatics GroupDepartment of Biochemistry Eotvos Lorand University, Budapest, Hungarydosztanyi@ceaser.elte.hu

Protein disorder is prevalent

20

40

60

0

LD

R (

40<

) p

rote

in, %

kingdom

B

A

E

Protein disorder is functional

Xie H et al. J Proteome Res. 2007, 6, 1882-1898

Protein disorder is important

Prion protein Prion disease

CFTR Cystic fibrosis

Alzheimer’s

-synuclein Parkinson’s

p53, BRCA1 cancer

Functions of intrinsically disordered proteins

I Entropic chains

II Linkers

III Molecular recognition

IV Protein modifications (e.g. phosphorylation)

V Assembly of large multiprotein complexes

Interactions of IDPsComplex between p53 and MDM2

Coupled folding and binding

Interactions of proteins

Interactions of proteins

Coupled folding and binding

Functional advantages

Weak transient, yet specific interactions Post-translational modifications Flexible binding regions that can overlap Evolutionary plasticity

SignalingRegulation

Coupled folding and binding Experimental difficulties

Highly flexible Weak interactions Short half-life

Complexes of IDPs in the PDB: ~ 200 Known instances: ~ 2,000 Estimated number of such interactions in the

human proteome: ~ 100,000

Interactions of IDPs

F19, W23 and L26

p27

Cyclin-dependent kinase (Cdk) inhibitor, p27Kip1 (p27) Cell cycle regulation Binds to cdk-cyclin komplex and inhibitis their activity Fully disordered protein

p27

p27

Partial unfolding enables the phosphorylation of Tyr88, starting a series of signaling events that leads to the beginning of S phase.

Prediction of functional regions within IDPs

Disordered binding regions (ANCHOR) Linear motifs (ELM, SlimPred) Morfs (Morfpred)

Specific conservation pattern

Disordered protein complexes

Complex between p53 and MDM2

• Interaction sites are usually linear (consist of only 1 part)

• enrichment of interaction prone amino acids

Sequence

Binding sites

No need for structure, binding sites can be

predicted from sequence alone

Prediction of disordered binding regions – ANCHOR

What discriminates disordered binding regions? A cannot form enough favorable interactions with their

sequential environment It is favorable for them to interact with a globular protein

Based on simplified physical model Based on an energy estimation method using statistical

potentials Captures sequential context

ANCHOR

Human p53 C –terminal region

ANCHOR and linear motifsNCOA2 transcription co-activator

The regions between 600-800 is disordered Contains 3 receptor binding motifs: xLxxLLx (LIG_NRBOX)

LMs and Disordered Binding Regions

Linear motifs and disordered binding regions often overlap

Complementary approaches

Prediction of disordered binding regions can help to increase likelihood of true instances

LMs and Disordered Binding Regions

Linear motifs and disordered binding regions often overlap

Complementary approaches

Prediction of disordered binding regions can help to increase likelihood of true instances

Machine learning approaches

SlimPred: trained on ELM database

Morfpred: trained on short chains in complex

Very small datasets

Negative datasets

Conservation

Conservation patterns of linear motifs

No evolutionary constraints to keep the structure Strong constraints on functional site

Island-like conservation

SlimPrints

Generates sequence alignments of orthologous sequences

Relative conservation score per position Filters out less reliable regions Fails if sequences are too divergent, or too similar http://bioware.ucd.ie/slimprints.html.