1. Structural Biology: What does 3D tell us?
Stephen J Everse
University of Vermont

2. Outline Determining a 3D structure Structural elements
X-ray crystallography
Structural elements
Modeling a 3D structure

3. Protein Structures Primary Secondary Tertiary Quaternary Amino acid
sequence.
Alpha helices &
Beta sheets,
Loops.
Arrangement
of secondary
elements in 3D space.
Packing of several
polypeptide chains.
Given an amino acid sequence, we are interested in its secondary structures, and how they are arranged in higher structures.

4. Secondary Structure a Helix
First predicted by Linus Pauling. Modeled on basis x-ray data which provided accurate geometries, bond lengths, and angles. Modeled before Kendrew’s structure;
3.6 residues/ turn, 5.4Å/ turn;
The main chain forms a central cylinder with R-groups projecting out;
Variable lengths: from 4 to 40+ residues with the average helix length is 10 residues (3 turns).

5. Secondary Structure The b Sheet
Unlike a helix, b sheet composed of secondary structure elements distant in structure;
The b strands are located next to each other
Hydrogen bonds can form between C=O groups of one strand and NH groups of an adjacent strand.
Two different orientations
all strands run same direction: “parallel”
strands in alternating orientation: the “antiparallel”.

6. b-Turns
Type I: Also referred to as a b turn: H-bond between Acyl O of AA1 and NH of AA4;
Type II, glycine must occupy the AA3 position due to steric effects;
Type III is equivalent to 310 helix;
Types I & III constitute some 70% of all b turns;
Proline is typically found in the second position, and most b turns have Asp, Asn, or Gly at the third position.

7. Other Secondary Structural Elements
Random coil
Loop
-turn
defined for 3 residues i, i+1, i+2 if a hydrogen bond exists between residues i and i+2 and the phi and psi angles of residue i+1 fall within 40 degrees of one of the following 2 classes
turn type phi(i+1) psi(i+1)
classic
inverse
Disordered structure

8. Viewing Structures Ball-and-stick CPK Ca or CA
It’s often as important to decide what to omit as it is to decide what to include
What you omit depends on what you want to emphasize

9. Ribbon and Topology Diagrams Representations of Secondary Structures
a-helix
b-strand N

10. Tools for Viewing Structures
Jmol
PyMOL
Swiss PDB viewer
Mage/KiNG
Rasmol

11. RCSB

12. GRASP Graphical Representation and Analysis of Structural Properties
Red = negative surface charge
Blue = positive surface charge

13. Consurf
The ConSurf server enables the identification of functionally important regions on the surface of a protein or domain, of known three-dimensional (3D) structure, based on the phylogenetic relations between its close sequence homologues;
A multiple sequence alignment (MSA) is used to build a phylogenetic tree consistent with the MSA and calculates conservation scores with either an empirical Bayesian or the Maximum Likelihood method.

14. How do we show 3-D? Stereo pairs Dynamics: rotation of flat image
Rely on the way the brain processes left- and right-eye images
If we allow our eyes to go slightly wall-eyed or crossed, the image appears three-dimensional
Dynamics: rotation of flat image
Perspective

15. Stereo pair: Release factor 2/3 Klaholz et al, Nature (2004) 427:862

16. Movies

17. Proteopedia

18. Protein structures in the PDB
The last 15 years have witnessed an explosion in the number of known protein structures. How do we make sense of all this information?
blue bars: yearly total
red bars: cumulative total

19. Classification of Protein Structures
The explosion of protein structures has led to the development of hierarchical systems for comparing and classifying them.
Effective protein classification systems allow us to address several fundamental and important questions:
If two proteins have similar structures, are they related by common ancestry, or did they converge on a common theme from two different starting points?
How likely is that two proteins with similar structures have the same function?
Put another way, if I have experimental knowledge of, or can somehow predict, a protein’s structure, I can fit into known classification systems. How much do I then know about that protein? Do I know what other proteins it is homologous to? Do I know what its function is?

20. Definition of Domain
“A polypeptide or part of a polypeptide chain that can independently fold into a stable tertiary structure...”
from Introduction to Protein Structure, by Branden & Tooze
“Compact units within the folding pattern of a single chain that look as if they should have independent stability.”
from Introduction to Protein Architecture, by Lesk
Thus, domains:
can be built from structural motifs;
independently folding elements;
functional units;
separable by proteases.
note that Lesk’s definition is more careful...
sometimes domains within multidomain proteins can evolve to have dependent stabilities, even when the same types of units occur independently in other proteins.
Two domains of a bifunctional enzyme

21. Proteins Can Be Made From One or More Domains
Proteins often have a modular organization
Single polypeptide chain may be divisible into smaller independent units of tertiary structure called domains
Domains are the fundamental units of structure classification
Different domains in a protein are also often associated with different functions carried out by the protein, though some functions occur at the interface between domains
domain organization of P53 tumor suppressor
activation
domain
sequence-specific
DNA binding domain
non-specific
DNA-binding
domain
tetramer-
ization
domain

22. Rates of Change Why? Not all proteins change at the same rate;
Functional pressures
Surface residues are observed to change most frequently;
Interior less frequently;

23. SequenceStructureFunction
Many sequences can give same structure
Side chain pattern more important than sequence
When homology is high (>50%), likely to have same structure and function (Structural Genomics)
Cores conserved
Surfaces and loops more variable
*3-D shape more conserved than sequence*
*There are a limited number of structural frameworks*
W. Chazin © 2003

24. Degree of Evolutionary Conservation
Less conserved
Information poor
More conserved
Information rich
DNA seq
Protein seq
Structure
Function
ACAGTTACAC
CGGCTATGTA
CTATACTTTG
HDSFKLPVMS
KFDWEMFKPC
GKFLDSGKLG
S. Lovell © 2002

25. How is a 3D structure determined ?
1. Experimental methods (Best approach):
X-rays crystallography - stable fold, good quality crystals.
NMR - stable fold, not suitable for large molecule.
2. In-silico methods (partial solutions -
based on similarity):
Sequence or profile alignment - uses similar sequences,
limited use of 3D information.
Threading - needs 3D structure, combinatorial complexity.
Ab-initio structure prediction - not always successful.

26. Experimental Determination of Atomic Resolution Structures
X-ray
X-rays
Diffraction
Pattern
Direct detection of
atom positions
Crystals
NMR RF
Resonance H0
Indirect detection of
H-H distances
In solution

27. • • Resolving Power d Resolving Power:
Signal
Position
Resolving Power:
The ability to see two points that are separated by a given distance as distinct
Resolution of two points separated by a distance d requires radiation with a wavelength on the order of d or shorter:
wavelength
Mark Rould © 2007

28. X-ray Microscopes? nair nair nglass ∆ There are no lenses for xrays.
•Lenses require a difference in refractive index between the air and lens material in order to 'bend' and redirect light (or any other form of electromagnetic radiation.)
•The refractive index for x-rays is almost exactly 1.00 for all materials.
∆ There are no lenses for xrays.
How do lenses work?
We'll see how we deal with this minor difficulty in a few slides.
There's another problem with trying to obtain atomic resolution images of specimens with x-rays...
Mark Rould © 2007

29. Fourier Transform of specimen
Light Scattering and Lenses are
Described by Fourier Transforms
Scattering =
Fourier Transform of specimen
Lens applies a second Fourier Transform to the scattered rays to give the image
There are no lenses for xrays.
The mathematical equivalent of a Lens = Fourier Transform.
Scattering = Fourier Transform of specimen
Lens applies a second Fourier Transform to the scattered rays to give the image
Since X-rays cannot be focused by lenses and refractive
index of X-rays in all materials is very close to 1.0 how do we get an atomic image?
Mark Rould © 2007

30. X-ray Diffraction with “The Fourier Duck”
The molecule
The diffraction pattern
Images by Kevin Cowtan

31. Animal Magic The CAT (molecule) The diffraction pattern
Images by Kevin Cowtan

32. Solution: Measure Scattered Rays, Use Fourier Transform to Mimic Lens Transforms
X-Ray Detector
Computer
Instead of using a lens to recombine the scattered ray, we capture the scattered x-rays with an x-ray detector.
To get an image of the specimen, just take the inverse Fourier transform of the diffraction data, right?
No -- we only have half the data that are needed to mimic a lens!
Mark Rould © 2007

33. A Problem…
A single molecule is a very weak scatterer of X-rays. Most of the X-rays will pass through the molecule without being diffracted. Those rays which are diffracted are too weak to be detected.
Solution: Analyzing diffraction from crystals instead of single molecules. A crystal is made of a three-dimensional repeat of ordered molecules (1014) whose signals reinforce each other. The resulting diffracted rays are strong enough to be detected.
3D repeating lattice;
Unit cell is the smallest unit of the lattice;
Come in all shapes and sizes.
Crystals come from slowly precipitating the biological molecule out of solution under conditions that will not damage or denature it (sometimes).
A Crystal
Sylvie Doublié © 2000

34. Putting it all together: X-ray diffraction
Electron
density map
Rubisco diffraction pattern
Crystallographer
Detector
Computer
Scattered rays
Object
X-rays
Diffraction pattern is a collection of
diffraction spots (reflections)
Model
Sylvie Doublié © 2000

35. What information does structure give you?
3-D view of macromolecules at near atomic resolution.
The result of a successful structural project is a “structure” or model of the macromolecule in the crystal.
You can assign:
- secondary structure elements
- position and conformation of side chains
- position of ligands, inhibitors, metals etc.
A model allows you:
- to understand biochemical and genetic data
(i.e., structural basis of functional changes in mutant
or modified macromolecule).
- generate hypotheses regarding the roles of particular
residues or domains
Sylvie Doublié © 2000

36. What did I just say????!!! A structure is a “MODEL”!!
What does that mean?
It is someone’s interpretation of the primary data!!!

37. So what happens when we can’t get an NMR or X-ray structure?

38. 2˚ & 3˚ Structure Prediction38 38

39. Secondary (2o) Structure
Table 10 -

40. Secondary Structure Prediction
One of the first fields to emerge in bioinformatics (~1967)
Grew from a simple observation that certain amino acids or combinations of amino acids seemed to prefer to be in certain secondary structures
Subject of hundreds of papers and dozens of books, many methods…

41. Simplified C-F Algorithm
Select a window of 7 residues
Calculate average Pa over this window and assign that value to the central residue
Repeat the calculation for Pb and Pc
Slide the window down one residue and repeat until sequence is complete
Analyze resulting “plot” and assign secondary structure (H, B, C) for each residue to highest value

42. Protein Principles Proteins reflect millions of years of evolution.
Most proteins belong to large evolutionary families.
3D structure is better conserved than sequence during evolution.
Similarities between sequences or between structures may reveal information about shared biological functions of a protein family.

43. The PhD Algorithm
Search the SWISS-PROT database and select high scoring homologues
Create a sequence “profile” from the resulting multiple alignment
Include global sequence info in the profile
Input the profile into a trained two-layer neural network to predict the structure and to “clean-up” the prediction

44. Best of the Best PredictProtein-PHD (72%)
Jpred (73-75%)
SAM-T08 (75%)
PSIpred (77%)

45. Structure Prediction T Threading
A protein fold recognition technique that involves incrementally replacing the sequence of a known protein structure with a query sequence of unknown structure.
Why threading?
Secondary structure is more conserved than primary structure
Tertiary structure is more conserved than secondary structure T H R E A D 45

46. 3D Threading Servers
Generate 3D models or coordinates of possible models based on input sequence
PredictProtein-PHDacc
PredAcc
bin/portal.py?form=PredAcc
Loopp (version 2)
Phyre
SwissModel
All require addresses since the process may take hours to complete

47. Ab Initio Folding Two Central Problems
Sampling conformational space (10100)
The energy minimum problem
The Sampling Problem (Solutions)
Lattice models, off-lattice models, simplified chain methods, parallelism
The Energy Problem (Solutions)
Threading energies, packing assessment, topology assessment

48. Lattice Folding

49.
Critical Assessment of protein Structure Prediction (CASP)

51. For the gamers out there…

52. Print & Online Resources
Crystallography Made Crystal Clear, by Gale Rhodes
Online tutorial with interactive applets and quizzes.
Nice pictures demonstrating Fourier transforms
Cool movies demonstrating key points about diffraction, resolution, data quality, and refinement.
Notes from a macromolecular crystallography course taught in Cambridge

54. Evolutionarily Conserved Domains
Often certain structural themes (domains) repeat themselves, but not always in proteins that have similar biological functions.
This phenomenon of repeating structures is consistent with the notion that the proteins are genetically related, and that they arose from one another or from a common ancestor.
In looking at the amino acid sequences, sometimes there are obvious homologies, and you could predict that the 3-D structures would be similar. But sometimes virtually identical 3-D structures have no sequence similarities at all!

55. The Motif
There are certain favored arrangements of multiple secondary structure elements that recur again and again in proteins--these are known as motifs or supersecondary structures
A motif is usually smaller than a domain but can encompass an entire domain. Sometimes the structures of domains are partly named after motifs that they contain, e.g. “greek key beta barrel”
It should be noted that the term motif, when used in conjunction with proteins, sometimes also refers to sequence features with an associated function, e.g. the “copper binding motif” HXXXXH.
motif is a fairly broad concept
a motif is usually part of a domain and does not fold on its own: for instance, some domains are composed of repeating beta-alpha-beta units.
but sometimes a motif is big enough to fold on its own and encompasses a whole domain--a jellyroll barrel might be an example of that.
when using the word motif to describe proteins, it sometimes has a functional rather than a structural meaning, like a phosphate-binding motif.
here, though, we are speaking in a purely structural sense.
“greek key” motif
beta-alpha-beta motif

56. Limitations of Chou-Fasman
Does not take into account long range information (>3 residues away)
Does not take into account sequence content or probable structure class
Assumes simple additive probability (not true in nature)
Does not include related sequences or alignments in prediction process
Only about 55% accurate (on good days)

57. Prediction Performance

58. An Approach SAS Calculations
DSSP - Database of Secondary Structures for Proteins
VADAR - Volume Area Dihedral Angle Reporter
GetArea
Naccess - Atomic Solvent Accessible Area Calculations