Protein-protein interactions

1
Protein-protein interactions

• Marcotte EM, Pellegrini M, Ng HL, Rice DW,
  Yeates TO, Eisenberg D. (1999). Detecting protein
  function and protein-protein interactions from
  genome sequences. Science 285, 751-3
• Enright AJ, Iliopoulos I, Kyrpides NC, Ouzounis
  CA. (1999). Protein interaction maps for complete
  genomes based on gene fusion events. Nature 402,
  86-90
• compare briefly with yeast-2-hybrid system (y2h)

2
Rosetta stone sequences

• protein A is homologous to subsequence from
  protein C
• protein B is homologous to subsequence from
  protein C
• subsequences from A and B are NOT homologous to
  each other

3
Rationale

• Proteins A and B form a multisubunit complex
  which is fused into a single protein in sequence
  C
• thermodynamics- pieces dont need to find each
  other in cell
• efficiency- cell needs to produce much less of
  each as a result
• metabolic channeling- mentioned in Nature
  papers last paragraph
• - enzymes in related biochemical pathway may form
  functional complexes
• substrates could then pass from one enzyme to
  another directly, instead of diffusing into the
  cytosol at large
• not clear if there is direct evidence showing
  metabolic channeling anywhere- (tryptophan
  synthase?)

4
Marcotte, et al (July 1999)

• Method 1 use domain subsequences defined by
  Pro-Dom
• all pairs of subsequence matches considered and
  searched for
• two proteins which have one from each pair
  matched against
• Method 2 sequence comparison
• two non-overlapping local sequence alignments to
  a third protein
• both use a minimal threshold for iding
  statistical significant scores

5
Marcotte, et al (July 1999) (2)

• Trying to test accuracy of independent
  predictions
• Method 1 shared keywords in SWISS-PROT
  annotations
• - golly gee its better than random
• - 68 vs 15 in E. coli
• - 32 vs 15 in S. cerevisiae (yeast)
• Method 2 Database of interacting Proteins
• 6.4 of applicable sequences are also in the
  database
• Rosetta Stone is not comprehensive (more on this
  later)
• Method 3 phylogenetic profiles (see last weeks
  papers)
• wow its better than random too
• 5, eight times as many as random

6
Enright, et al (NOV 1999)

• use BLASTP to compare query genome against
  itself
• formation of a binary matrix (1s or 0s in each
  entry)
• - symmetrification with Smith-Waterman (local
  alignments
• BLASTP to compare query vs reference genome
• get a second binary matrix
• all pairs of query proteins similar to a
  reference protein
• Z-score to test for significance of alignments
• - set an arbitrary Z-score cutoff to determine
  coverage/accuracy

7
And now for something completely different

• y2h gt yeast-2-hybrid detection of
  protein-protein interactions
• - Fields, S. and Song, O. Nature 1989, if youre
  curious.
• transcription factors composed of two separable
  domains
• - DNA-binding and transcriptional-complex
  recruitment

8
High throughput y2h

• is only high-throughput if you are not a
  postdoc.
• lots of transformations and assays
• fortunately you only have to transform once.
• has major problems with false negatives
• integral membrane proteins dont work (dont
  fold properly)
• post-translational modifications
• require nuclear localization
• misfolding or steric hindrances
• transcription factors (?)
• also has significant false positives also- not
  sure why
• - Uetz et al (2000) had 20 of interactions
  screen twice
• other validation methods
• genetic techniques
• biochemical (coIP, affinity chromatography, mass
  spec, etc)

9
A brief discussion on signal transduction

• kinase cascades are ubiquitous in signaling
  pathways
• MKKK gt MKK gt MAPK
• kinase cascades are ubiquitous in signaling
  pathways
• interesting when looking at given Rosetta Stone
  examples
• will kinase cascades be detected?
• SH2 and SH3 domains (Marcotte, et al)
• SH2 bind phosphorylated tyrosine residues, SH3
  bind proline-rich sequences
• both are common motifs but have
  sequence-specific affinity
• Kinase cascades/signaling pathways are sometimes
  Y2H targets