Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out

intrinsic disorder in cell signaling and cancer patients

VIEWS: 60 PAGES: 33

									                Intrinsic Disorder in
                    Cell-signaling and
Cancer-associated Proteins

Center For Computational
Biology and Bioinformatics
                Outline of Talk
•   Recognition of intrinsic disorder by example
•   Prediction of natural disordered regions
•   Disorder and cell signaling
•   Disorder and cancer
•   Disorder and drug discovery
                                           Calcineurin

                                                 B-Subunit




                                                                               A-Subunit




                                                      Active Site




                                                              Autoinhibitory
                                                                 Peptide


Kissinger C. et al., Nature 378:641-644 (1995)
             Calmodulin Binding to Target




Meador W. et al., Science 257: 1251-1255 (1992)
                        Disorder and Function
      Category                 Change Examples                   Descriptions
      Molecular                  DO                  113      Inter- and Intra-protein,
      Recognition                                           ssDNA, dsDNA, tRNA, rRNA,
                                                               mRNA, nRNA, bilayers,
                                                             ligands, co-factors, metals
      Protein                   Variable              36     Acetylation, fatty acylation,
      Modification                                           glycosylation, methylation,
                                                                phosphorylation, ADP-
                                                             ribosylation, ubiquitination,
                                                                 proteolytic digestion
      Entropic                 Variable               17     Linkers, spacers, bristles,
      Chains                                                clocks, springs, detergents,
                                                                   self-transport

Dunker AK et al., Adv Protein Chem 62: 25-49 (2002)
       Disorder and Molecular Recognition:
              The Schulz Diagram
   a                       b



  a’                      b’


                                                                         b’



Schulz G., Molecular Mechanisms of Biological Recognition 79-94 (1979)
           Disorder and
       Molecular Recognition
                     STRUCTUAL
CHARACTERISTIC       VARIABILITY
High Specificity /   Flexibility in the
Low Affinity         Unbound State
Reversibility /      Polymorphism in the
Multiple Specificity Bound State
   Glycogen Synthase Kinase 3β
• One GSK3b site exhibits mutually exclusive
  binding to Frat & Axin

• The structure of the GSK3b site is ill-defined w/o
  partner, but…

• This site adopts different structures in a partner-
  dependent manner (dimorphism in the bound
  state)
     Glycogen Synthase Kinase 3β (GSK3β)




       Axin (382-401)     Frat (197-222)   GSK3β-Axin   GSK3β-Frat


       Axin peptide:           VEPQKFAEELIHRLEAVQ
         FRAT peptide: SQPETRTGDDDPHRLLQQLVLSGNLIKEAVRRLHSRRLQ


Dajani et al., EMBO J. 22:494-501 (2003)
     Prediction of Disorder
       Disordered Sequence Data

     Attribute Selection or Extraction

    Separate Training and Testing Sets

            Predictor Training

Predictor Validation on Out-of-Sample Data

                Prediction
                                        Amino Acid Compositions
                                          1
           (Disorder – Order) / Order


                                                 dis XRAY (2844)
                                                 dis NMR (4019)
                                        0.5
                                                 dis CD (10554)


                                          0




                                        -0.5




                                         -1
                                               W C   F   I   Y   V   L   H M A   T   R G Q   S   N   P   D   E   K

Dunker AK et al., Adv Protein Chem 62: 25-49 (2002)
              www.disprot.org
DisEMBLTM    Intrinsic Protein Disorder Prediction
DISOPRED2    Disorder Prediction Server
DRIPPRED     Web based predictor for disordered regions in proteins
FoldIndex©   Estimate the fold probability of a protein
             Intrinsic Protein Disorder, Domain & Globularity
GlobPlot 2
             Prediction
IUPred       Prediction of Intrinsically Unstructured Proteins
PONDR®       Predictors of Natural Disordered Regions
             Prediction of unfolded segments in a protein sequence
PreLink
             based on amino acid composition
RONN         Regional Order Neural Network
VL2          DisProt Predictor of Intrinsically Disordered Regions
VL3, VL3H,
             DisProt Predictor of Intrinsically Disordered Regions
VL3E
   Predictors of Naturally Disordered
          Regions (PONDR®S)
• Inputs: typically amino acid compositions,
  sequence complexity, hydropathy, net charge,
  etc. over windows of various lengths from 21 to 41
• VL-XT: Three Neural Networks, one for internal
  residues and one for each terminus
• VL2: Linear Least Squares, special for termini
• VL3: Neural Network Ensembles
• VSL1 (ISTZORAN) – Three predictors, two
  stages: one for short regions, one for long, one to
  integrate these
ROC Curves
                Protein Interaction Domains




http://www.mshri.on.ca/pawson/domains.html
      Crystal Structures & Protein-Protein
                  Interactions




Freund et al., (2002) Embo J. 21:5985-5995
                          CD2-Binding Protein




Freund et al., (1999) Nat. Struct. Biol. 6:656-660
     CD2-Binding partner to CDBP2 GYF
                  domain




         Consensus sequence (GYF binding sites) has the sequence: ppppghr. The
            peptide in the crystal structure has the aa sequence: shrppppghrv.
Freund et al., (1999) Nat. Struct. Biol. 6:656-660
   Analysis of Signaling Interactions
• Examined 44 interactions on Pawson’s website.

• Almost all of the interactions involved ordered
  regions binding to disordered partners.

• Conclusion: if Pawson’s examples are typical,
  then almost all signaling interactions use
  disordered proteins.
Molecular Recognition Feature (MoRF)
                                 4E – Binding Protein
                  1
Relative Score




                 0.5




                  0
                       0            20           40           60               80         100
                                                        Residue
                           EF4E Binding Region        PONDR        PHD Helix        Hydrophobic Moment
       RNase E Organizing Domain
                                       A B          C   D      Binding Partners
                                                            A Self Association
                                                            B Arg-rich, coiled-coil
                                                               RNA binding region
                                                            C Enolase: structure of
                                                               the complex shows a
                                                               helix for the fragment
                                                            D Polynucleotide
                                                               phosphorylase


Callaghan A et al., J Mol Biol 340:965-979 (2004)
      Predictors of a-forming MoRFs
• Training set: 14 a-MoRFs versus 1,200 globular,
  ordered proteins – both from PDB
• Inputs: short predictions of order, flanked by
  predictions of disorder using PONDR® VL-XT;
  flanking regions exhibit absence of hydrophobic
  clusters, disorder by VL2, low hydrophobic moment
  values, and GOR I prediction of coil and turns
• Adjust thresholds to reduce false positive error rate
  on helices from globular proteins while avoiding loss
  of training set examples (In progress)
             α-MoRF Predictions
            Across Three Kingdoms
                  % of Proteins with Predicted MoRFs
             0        5       10       15         20    25


Eukaryotes


 Bacteria
                                    % of Proteins

 Archaea                            MoRFs/Residue

            0.0      0.2      0.4     0.6        0.8    1.0
                    Predicted MoRFs/residue   (x10-3)
        Predicted MoRF of Measles Virus




                                                 488-499



Bourhis J et al., Virus Res 99: 157-167 (2004)
      Measles Virus N and P Proteins


       Predicted a-MoRF:
       Residues 488 to 499.

       Observed a-MoRF:
       Residues 486 to 504




Kingston R et al., Proc Nal Acad Sci USA 101: 8301-8306 (2004)
                           Disorder and Cell Signaling
                          80
                                                            Cancer-associated proteins
                                                            Signaling proteins
                                                            SwissProt
                          60                                O_PDB Select25
          % of proteins




                          40



                          20



                          0
                               >=30   >=40   >=50    >=60   >=70   >=80   >=90 >=100
                                        consecutive disorder predictions
Iakoucheva I et al., J Mol Biol 323:573-584 (2002)
p53Disordered Binding Sites
  PONDR® VL-XT
     Score
      Disorder and Drug Discovery
• The p53-MDM2 interaction is blocked by several drugs; one
  is in clinical trials and shows promise as an anti-cancer drug.
• The drug molecules bind to the ordered partner, preventing
  the disordered partner from binding.
• Such interactions are typically weak per unit of surface area,
  and the interaction surfaces can be small, thus such
  interactions are ideal drug targets.
• Molecular Kinetics has strategy to find all druggable MoRF-
  based interactions; bioinformatics indicates that more than
  one hundred are in cancer-associated proteins.
• Is this approach a new drug discovery pathyway?
    Parallel Paradigms
           Catalysis
AA seq → 3-D Structure → Function


           Signaling
  AA seq → Disordered → Function
           Ensemble
                Thanks!
    kedunker@iupui.edu

    Subject: Reprint Request
.

    Database: http://www.disprot.org

    Predictions: http://www.PONDR.com
                  Support
•   NSF CSE II 9711532
•   NIH R01 LM007688-01A1
•   USDA 2000 1740
•   INGEN®, Lilly Endowment
•   Molecular Kinetics
                Acknowledgements
  Disorder Lab        Temple University      Indiana University
   Vladimir Uversky     Zoran Obradovic        Predrag Radivojac
     Marc Cortese       Slobodan Vucetic         Pedro Romero
   Shelley Riggen        Vladimir Vacic
   Andrew Campen           Kang Peng
     Ya-Yin Fang
                                           Rockefeller University
                                              Lilia Iakoucheva Sebat
      Gerard Go             UCSF
    Amrita Mohan          Ethan Garner
       Jie Sun                             University of Wisconsin
     Siama Zaida                                  Chris Oldfield
     Yizhi Zhang            PNNL
                         Richard Smith
                         Eric Ackerman
                                             Molecular Kinetics
University of Idaho                                Ya-Yue Van
   Celeste J. Brown                              Tanguy LeGall
    Chris Williams                               Yugong Cheng
                                                  Jessica Chen
        Jessica Chen

								
To top