Annotation of the Human Genome by HighThroughput Sequence Analysis of Naturally Occurring Proteins

1
Annotation of the Human Genome by High-Throughput
Sequence Analysis of Naturally Occurring Proteins

• Christopher Southan PhD
• Principle Scientist, Molecular Pharmacology,
  AstraZeneca RD, Mölndal, Sweden
• Special Professor of Proteomics, School of
  Biosciences, University of Nottingham

• Presentation Outline
• Issues with annotating human protein-coding genes
• Outline of the Oxford Genome Anatomy Project
• Examples of proteins with new biochemical
  information or inferences
• Conclusions

2
Lack of a Consensus Human ORFomeDefining the
Need for MS-based Protein verification

• Estimates of protein number declining but
  diversity increasing
• False positives and false negatives in public
  data
• Undiscovered small proteins (SMORFs)
• 2,645 Ensembl novel proteins
• The International Protein Index indicates churn
  and non-overlap
• Alternative splice forms underrepresented
• Patent protein sequences not included in genome
  annotation
• Parallel annotation efforts
• Pseudogenes difficult to define
• Non-coding transcripts
• Unvalidated aa-changing SNPs

3
Oxford Genome Anatomy Project (OGAP)

• Developed by Oxford Genome Sciences with elements
  of Confirmant Human Protein Atlas
• MALDI-TOF of peptide pools from large scale 1D-
  and 2D gel plugs is followed by selected tandem
  MS/MS
• Peptides are identified using a maximal virtual
  tryptome search space to include unverified
  proteins, nsSNPs and splice forms
• Database constructed from protein sequences
  mapped to Golden Path and optimally fitted to
  the MS/MS and MALDI data
• Identified proteins are linked, via the MS data,
  to gel positions and samples
• Details at www.OGAP.co.uk

4
Data Mining, Links and Statistics

• 1 million MALDI determinations and 300K
  MS/MS spectra
• 13,000 MS/MS supported proteins
• Samples of 70 cell types derived from 50 tissues
• Over 20 sub-cellular fractions
• Merge of three identification pipelines (ICAT, 2D
  and 1D)
• Peptide assignments supported by multiple MS
  scoring methods
• Data anchored on Ensembl Golden Path

5
Verifying and Correcting Gene Predictions
Transmembrane Anchor Protein (TMAP)
Fractions Cells / tissues
Exons MS data

• Tandem MS/MS peptide sequences confirmed partial
  but incorrect gene predictions from NCBI
  XP_301613 and ENSP00000332787 on chromosome 17
• 1D-gel samples from endothelial and neuronal stem
  cell membrane fractions
• Two full-length splice forms assembled from EST
  and MS data
• ESTs suggest broad tissue distribution
• Sequence confirmed as partial cDNA HBV
  PreS1-transactivated protein 1
• No paralogues but many orthologues, no detectable
  structural homology

6
TMAP Bioinformatic Analysis of An Unknown
Orphan Protein
Pseudo signal peptide prediction
TM prediction indicates membrane anchor
Helical sections from 2D structure prediction
7
Napsin A Location in urine

• Observed 2D-gel Mr/pI consistent with removal of
  signal and activation peptide
• Observation supported by two independent
  publications using antibodies
• - Detection of immunoreactive napsin A in
  human urine. Schauer-Vukasinovic et al. BBA.
  2001, 1524(1)51-6.
• - Immunohistochemical localization of napsin
  and its potential role in protein catabolism in
  renal proximal tubules. Mori et al. Arch Histol
  Cytol. 2002, 65(4)359-68.

8
Cytokine FAMC3 Inference of Phosphorylation

• Four-helix-bundle cytokine without biochemical
  characterisation
• 1D-gel implies the protein is membrane associated
  (signal pep retention ?)
• 2D-gel migration supports signal removal and
  transition to the CSF
• 2D pI chain suggests 3 positions of
  phosphorylation

9
Acyl Peptide Hydrolase New Location

• Erthrocyte specific degradation of oxidatively
  damaged proteins
• Novel observations in CSF, brain tissues and
  cancer cells
• 2D-gel observed pI/Mr 5.2/80 kd
• Predicted pI/Mr 5.29/81.2
• Independent electrospray MS result Mr 81.2 kd

10
LOC146556 Verification of an Unknown

• Initially a high-throughput unannotated cDNA
• Subsequently submitted from the Genentech
  Secreted Protein Discovery Initiative
• No signal peptide but 60 aa N-terminal anchor
• Three splice variants in UniProt
• Novel observation of a CNS protein with Mw/pI
  suggestive of proteolytic clipping

11
Calsyntenin-1 Partitioning into CSF

• Calsyntenin-1 binds synaptic Ca2 thereby
  modulating Ca2-mediated postsynaptic signals
• May be subject to regulation by extracellular
  proteolysis in the synaptic cleft
• Multiple observations in CSF, probably after
  proteolysis
• 2D Spot spread suggestive of glycosylation

12
Limbic-system-associated Membrane Protein

• LAMP 64-68-kDa neuronal surface glycoprotein
  from limbic system
• Predicted GPI anchor and eight putative
  N-glycosylations
• Observed Mr and acidic shift on 2D-gel infers
  loss of the GPI anchor by transition to CSF and
  possible glycosylation

13
Conclusions

• Annotation of the human genome by the OGAP
  high-throughput MS protein verification pipeline
  can resolve some of the uncertainties of in slico
  annotation and contribute to closure of the basal
  proteome
• The expanding range of clinical samples, tissues,
  cell lines, sub-cellular fractions, data
  integration and mining tools produces a unique,
  high value database
• The combination of psedogel interpretation,
  sample knowlege and bioinformatic analysis can
  reveal novel details and inferences
• Many of these inferences are testable and can
  generate information that compliments public data
  for biochemicaly characterised and unknown
  proteins

14
References and Acknowledgments

• Rohlff (2004) New approaches towards integrated
  proteomics databases and depositories. Expert
  Reviews in Proteomics, 1 (3), 267 274 , review.
• McGowan, S et al. (84 authors) (2004) Annotation
  of the Human Genome by High-Throughput Sequence
  Analysis of Naturally Occurring Proteins,
  Current Proteomics, 1, 41-48
• Southan C. Has the Yo-yo stopped? a human gene
  number update (2004) Proteomics (6)1712-26,
  review, (also see poster)
• Rohlff Southan (2002) Proteomic approaches to
  central nervous system disorders, Curr Opin Mol
  Ther. 2002 4(3)251-8, review.
• http//oxfordgenomesciences.com/

Disclaimer While the speaker is
pleased to present material from participation in
this work before joining AstraZeneca this does
not constitute an endorsment of the Oxford Genome
Sciences Oxford Genome Anatomy Project by
AstraZeneca