This document summarizes a presentation on the evolution of symmetry in protein structures. It discusses how CE-Symm, a software tool, can accurately detect internal symmetry in proteins. The presentation finds that 18% of SCOP superfamilies exhibit symmetry, with β proteins having the highest percentage at 24.6%. It also discusses how identifying symmetric subdomains can help find "protodomains" that are building blocks of protein domains and have spread across evolution. The use of symmetry to study protein evolution is demonstrated through several examples.

1. The evolution of symmetry in protein
fold space
Presenter: Douglas Myers-Turnbull
undergraduate student, bioinformatics
Systematic detection of internal symmetry in proteins
using CE-Symm
Douglas Myers-Turnbulla, Spencer E. Blivenb, Peter W. Rosec, Zaid K.
Azidd, Philippe Youkharibachee, Philip E. Bournef,*, Andreas Prlicc,*
Journal of Molecular Biology, under second-pass review.

2. Symmetry is common
PDB IDs from left to right:
1VYM, 3HDP, 1G62, 1U6D, 3DDV

3. Why is symmetry so common?
 Enzymatic function

4. Why is symmetry so common?
 Enzymatic function
 Lowest energy state

5. Why is symmetry so common?
 Enzymatic function
 Lowest energy state
 Fewest kinetic barriers

6. Why is symmetry so common?
 Enzymatic function
 Lowest energy state
 Fewest kinetic barriers in folding
Easier to evolve complex structures from simple
building blocks

7. The evolution of symmetry in a β-trefoil

8. Building blocks of domains
Modular Evolution and the Origins of Symmetry:
“…symmetric protein structures can be constructed
from a set of basic ‘building blocks’ or subdomain
modules.”

9. What is a protodomain?
 A building block for domains

10. What is a protodomain?
 A building block for domains
 A subdomain that occurs across distant folds

11. What is a protodomain?
 A building block for domains
 A subdomain that occurs across distant folds
 Unlikely to have arisen by chance

12. Algorithms that detect symmetry
 sequence-based methods
Ex: DAVROS [Taylor et. al.]
 angle-based methods
Ex: Swelfe [Abraham et. al.]
 methods based on secondary structure
Ex: GANGSTA+ [Guerler et. al.]
 truly structural methods
Ex: SymD [Kim et. al.]

13. CE-Symm
 Align a structure against itself

14. CE-Symm
 Align a structure against itself
 Use Combinatorial Extension

15. A large benchmark set
 1,007 distinct SCOP superfamilies

16. A large benchmark set
 1,007 distinct SCOP superfamilies
 manually determined symmetry

17. CE-Symm is very accurate

18. Structural classification of proteins
 SCOP: class→fold→superfamily→family→domain

19. Structural classification of proteins
 SCOP: class→fold→superfamily→family→domain
 Different superfamilies of the same fold often have
substantial differences in structure

20. Normalization by superfamilies
 Normalize by number of domains per superfamily

21. Normalization by superfamilies
 Normalize by number of domains per superfamily
 A superfamily is symmetric if more than half of its
domains are symmetric.

22. A census of symmetry
SCOP class number of SFs % symmetric (SFs)
α 503 18.5%
β 354 24.6%
α/β 244 16.8%
α+β 549 14.3%
membrane 108 23.8%
overall 1824 18.0%
Complete results available at: http://source.rcsb.org

23. A census of symmetry
SCOP class number of SFs % symmetric (SFs)
α 503 18.5%
β 354 24.6%
α/β 244 16.8%
α+β 549 14.3%
membrane 108 23.8%
overall 1824 18.0%
Complete results available at: http://source.rcsb.org

24. A census of symmetry
SCOP class number of SFs % symmetric (SFs)
α 503 18.5%
β 354 24.6%
α/β 244 16.8%
α+β 549 14.3%
membrane 108 23.8%
overall 1824 18.0%
Complete results available at: http://source.rcsb.org

25. Symmetry and evolution
 How can we use this to learn about evolution?

26. Is this domain symmetric?
PDB ID: 3DDV. Zhang, R. et. al.

27. CE-Symm says it’s symmetric

28. Did we find a protodomain?
A protodomain?

29. Search for matching domains

30. A hit against another domain

31. It’s a β-propeller blade!
PDB ID: 1SHY. Stamos, J. et. al.

32. It’s a β-propeller blade!
PDB ID: 1SHY. Stamos, J. et. al.

33. Identifying protodomains systematically
1. Identify subdomains of symmetric structures with
CE-Symm

34. Identifying protodomains systematically
1. Identify subdomains of symmetric structures with
CE-Symm
2. Identify hits against other domains

35. Identifying protodomains systematically
1. Identify subdomains of symmetric structures with
CE-Symm
2. Identify hits against other domains
3. Derive a non-redundant set of protodomains

36. CE-Symm is on the web
http://source.rcsb.org

37. Acknowledgments
 Dr. Andreas Prlic, San Diego Supercomputer Center
 Spencer Bliven, Bioinformatics and Systems Biology
 Dr. Peter Rose, Skaggs School of Pharmacy
 Zaid Aziz, Chemistry and Biochemistry
 Dr. Philippe Youkharibache, Life Sciences R&D
 Dr. Phil Bourne, Skaggs School of Pharmacy

39. CE-Symm
1. Disable main diagonal (δ = 20)

40. CE-Symm
1. Disable main diagonal (δ = 20)
2. Duplicate matrix

41. CE-Symm
1. Disable main diagonal (δ = 20)
2. Duplicate matrix
3. Superimpose matrices

42. Current limitation: order-detection
 CE-Symm sometimes reports the wrong order

43. Two methods for order-detection
 Method 1: apply alignment repeatedly until the
composition becomes approximately the identity
 Method 2: identify the lowest difference ε(θ):

44. Most symmetric architectures are old

45. Types of symmetry in the benchmark

46. Some functions are related to symmetry

47. CE-Symm identifies symmetric folds
id fold CE-Symm (%) SymD (%) GANGSTA (%)
d.58 Ferredoxin-like 72 19 23
b.1 Immunoglobulin-like 61 8.9 8.4
b.42 Beta-Trefoil 97 100 56
a.24 Four-helical bundle 73 51 25
d.131 DNA clamp 100 91 64
b.69 7-bladed beta propeller 100 100 37
c.1 TIM beta/alpha barrel 87 83 3.7
b.11 Gamma-Crystallin-like 92 75 83

48. A stable synthetic β-trefoil
 Native function was lost

49. A synthetic adiponectin trimer
 Native function increased!

50. Symmetry is commonplace
 Quaternary symmetry
 Symmetry around active sites
PDB ID: 3HDP. Satyanarayana, L et. al.

51. Symmetry is commonplace
 Quaternary symmetry
 Symmetry around active sites
PDB ID: 2GZL. Crane, C.M. et. al.

52. Symmetry is commonplace
 Quaternary symmetry
 Symmetry around active sites
 Nested symmetry
PDB ID: 2GG6. Funke, T. et. al.

53. Symmetry is commonplace
 Quaternary symmetry
 Symmetry around active sites
 Nested symmetry
PDB ID: 2GG6. Funke, T. et. al.