Protein structure from X-ray diffraction Diffraction data quality

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

Diffraction images: Because of crystal symmetry and Friedels law, the
Protein, chemical structure:
IALEFGPSLKMNE measurement of diffraction data is redundant (multiple
measurements)
Reduncancy allows images to be scaled relative to each
Conformation, 3D-structure: other, even when X-ray intensity or crystal volume
CRYST1 221.200
ATOM 1 N
73.600
ILE A 6
80.900 90.00 90.00
97.764 18.390
90.00 P 21 21 2
39.211 1.00 84.23
4 irradiated varies from image to image (other more
ATOM
ATOM
ATOM
2 CA
3 C
4 O
ILE A
ILE A
ILE A
6
6
6
97.130 18.979
96.655 17.885
97.460 17.052
37.983 1.00 84.74
37.031 1.00 84.98
36.605 1.00 85.82
complicated effects: radiation damage, absorbtion of
ATOM
ATOM
ATOM
5 CB
6 CG1
7 CG2
ILE A
ILE A
ILE A
6
6
6
98.139 19.855
99.043 18.979
98.984 20.617
37.248 1.00 99.99
36.389 1.00 99.99
38.263 1.00 99.99
crystal and air)
Multiple measurements are averaged, giving more accurate
ATOM 8 CD1 ILE A 6 100.297 18.614 37.178 1.00 99.99
ATOM 9 N ALA A 7 95.359 17.896 36.702 1.00 84.56
ATOM 10 CA ALA A 7 94.757 16.906 35.795 1.00 83.75

values and an estimate of the statistical error

ATOM 11 C ALA A 7 95.816 16.303 34.876 1.00 83.47
.
.
.
ATOM 7604 O2* G R 3 73.810 43.517 21.774 1.00 62.48
Parameters that are quoted in the literature: Signal/noise
Reciprocal space (or Rmerge), completeness, redundancy
Real space

Crystal symmetry Crystal Symmetry Operations

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

Symmetry operations Crystallographic symmetry operations are valid
Unit cell and asymmetric unit throughout the crystal
Symmetry elements E.g. rotation by 180 (half turn) around
Exercise: Fishes in different shapes and colors

:-{

:-{

:-{
:-{

:-{

:-{
Symmetry of reciprocal space
:-{

:-{

:-{
Friedel's law
:-{

:-{

:-{
:-{

:-{

:-{
:-{

:-{

:-{

Space group: the combined

Symmetry Elements
symmetry elements
X-ray crystallography course 2006, Karsten Theis, UMass Amherst

Example for symmetry element: 4-fold axis a 2-fold axis (1 2) and a lattice translation (23) X-ray crystallography course 2006, Karsten Theis, UMass Amherst

Includes the following operations: combined result in a 2-fold axis (1 3)

Rotation around the axis by 90, 180, 270 degrees
:-{

:-{
:-{

:-{

The following symmetry elements 2 1 3

are possible for crystals with chiral
content like proteins: 1/3 a 31 screw axis There are 230 different space groups describing
2,3,4,6-fold rotation axes along
101 a
3D symmetry
21,31/2,41/2/3/,61/2/3/4/5-fold screw axes
Translations a/2, b/2, c/2 in different Protein crystals are chiral; a mirror symmetry
combinations 1/3 would change chirality; this limits the number of
spacegroups to ~70 (so-called acentric ones)
 The asymmetric unit The content of the asymmetric unit

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

Crystal symmetry duplicates/triplicates etc. Matthews coefficient and solvent content
molecules placed into the unit cell VM=Volume(AU) / Mw (protein in AU)
The unique volume of the unit cell not Observed VM values typically are in the range of 2.0 .. 4.0 3/Da
related by crystal symmetry is called If we assume an average protein density (1.23 g/ml), we can calculate
asymmetric unit the solvent content from Vm
Non-crystallographic symmetry (NCS) Make assumption of how much protein is in AU and check Vm
the asymmetric unit often contains more than Known subunit structure vs. crystal symmetry
one copy of the protein
e.g. hexagonal or trigonal crystals of hexameric helicase
the operations superimposing these multiple
copies are called non-crystallographic symmetry
operations
NCS-related molecules arent identical and have
different crystal environments

Symmetry of the diffraction data Friedels law

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

If the crystal obeys a certain rotation symmetry, the
intensities in reciprocal space will as well. Screw axes in 2 4 4
the crystal also result in rotational symmetry in reciprocal q Fq
3
space. 3
kin kout 2 1
1

|F|HKL=|F|HKL

2 4
kout kin 2 1
We can distinguish between the presence of pure rotation 3
axes and screw axes in the crystal by so-called systematic 3
absences of axis reflections. Axis reflections have Miller -q
1 4 F-q
indices h00 (along a*), 0k0 (along b*), and 00l (along c*).

Structure Factors Fhkl

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

For each reflection, the structure factor Fhkl gives

the magnitude and the phase of the wave resulting How do we obtain
from interference of all atoms in the unit cell
The measured intensities Ihkl are proportional to
the crystal structure of a protein
the square of the structure factor magnitude |F|hkl from the diffraction data |Fhkl|?
The structure factors give the information about
the structure of whatever is inside the unit cell
 : Fourier transform Fourier tour in two dimensions

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

Diffraction: real space vs. reciprocal space (r) F(q) Light/dark:
Intensities

Colors:
Phases

pulsed NMR, FT-IR and other spectroscopic methods: Molecule Fourier transform
time series vs. frequencies e--density
2 1

1.5
0.8
1

real space reciprocal space

Amplitude

0.6
0.5
Signal

0 0.4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
-0.5
0.2
-1

-1.5 0
0 2 4 6 8 10 12
-2
time/sec frequency/Hz

Crystals amplify the scattering signal The Fourier transform is reversible

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

(r) F(h,k,l)

Fourier
Crystal transform

Fourier analysis Fourier synthesis

The signal from a single molecule would be much too weak to detect

The signals originating from the molecules in the crystal add up because the
molecules are in identical orientation

Scattering by a crystal (called diffraction) results in a pattern with discrete spots

and empty areas (another level of signal/noise increase)

Downside: diffraction image represents average structure (over time and crystal)
The Fourier duck lives at www.ysbl.york.ac.uk/~cowtan/fourier/fourier.html

Problem: without phases,

Why is solving and interpreting no electron density
a crystal structure difficult? Fourier F(q) Fourier X-ray crystallography course 2006, Karsten Theis, UMass Amherst

analysis synthesis

The Phase Problem

Crystals arent perfect because proteins are flexible

measured
Unknown structure diffraction pattern
Protein conformations differ to some extent depending on |F(q)| Fourier
their environment synthesis without phases:
gibberish
 Solving the phase problem Calculating phases from an atomic model

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

1. Direct methods (guessing the phases)
Works for small molecules (lots of measurements per atom in
structure) |Fo(hkl)| derived from
observed intensities
2. Patterson methods (inter-atomic vectors)
Works for simple structures (4 atom pairs with 2 atoms, 100 atom unknown structure measured
pairs with 10 atoms diffraction pattern

3. Model phases; molecular replacement

Relies on partial knowledge of the structure
2Fo-Fc map
4. MIR/MAD techniques (ab initio) |Fc(hkl)| and phases
Prepare crystals that contain Se, Hg, Pt or other heavy atoms at a derived from model
handful of positions in the crystal and solve that simple structure
first using method 1. or 2. atomic model of calculated
known structure diffraction pattern

Molecular replacement: unknown structure

has a distinct crystal packing MIR/MAD method
X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

Intensities Solve a simpler problem first: just a couple of atoms
Collect data of crystals multiple times, with slight variations

MAD MIR
Multiwavelength
measure identical crystal Multiple soak crystals in different
Unknown structure measured 2Fo-Fc map Anomalous at 3 different wavelengths Isomorphous substances and measure
diffraction pattern Dispersion near Se absorption edge Replacement
Phases
:-{

:-{

:-{

:-{

:-{

:-{
Se scattering is influenced by Pt derivative Hg derivative
choice of wavelength, scattering native
Known structure rotated structure calculated by other atoms is not
diffraction pattern

What is the best method? Resolution: real space

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

Lower resolution Higher resolution X-ray crystallography course 2006, Karsten Theis, UMass Amherst
MIR/MAD works for all protein structures
but
requires additional measurements, experience and
lots of work to place atoms into density

Molecular replacement is fast explanation coming up..

but
large model bias in low-resolution structures
errors are propagated
new features are overlooked

File size: 2 Kb 11 Kb 672 Kb

 Resolution: reciprocal space What prevents high resolution?

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

electrons have distribution around atom rather than one
single location (same for all crystals)
molecules move as rigid bodies in crystal packing (depends
on crystal contacts)
atom positions change with time (local dynamic disorder)
atom positions change from one unit cell to another (local
3 static disorder)

12 4 B-factor Displacement
Temperature factors (B-factors) 20 2 0.25
6
are introduced to account for 40 2 0.51
disorder in the atomic model
80 2 1.01

Low resolution data: loss of detail Low resolution leads to model bias
X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

weak signal

strong signal
no tail Intensities

Real structure measured

Fourier Fourier diffraction pattern
analysis synthesis
tail?

Phases Although the cat has

no tail in the real
structure, it appears
tail in the model-biased
Current model calculated density

diffraction pattern

B-factors and disorder Atomic displacement and B-factors

Synonyms: Temperature factors, atomic (atomic position) = sqrt [B-factor / (8 2)]
displacement factors
Describe how the electron density of an atom is
broadened by static and dynamic disorder in the B-factor in 2 Atomic displacement in
crystal
5 0.25
Static disorder: distinct atomic positions in
different unit cells of the crystal 20 0.50
Dynamic disorder: changes in conformation over
time during the measurement 50 0.80

80 1.01
 Disorder makes interpretation Disorder makes interpretation
more difficult more difficult
Shading
represents
noise due to
Phosphorous 1) errors in
experimental Two carbon atoms, 1.4 apart,
data B = 10 or 50 2 at a resolution of 0.7
2) errors in
phases derived
Carbon from model
B

Disorder makes data collection Electron density

(and solving a structure) more difficult

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

The intensity of the
diffraction data
5 2.5 1.7 1.2 1 decreases with
resolution
sqrt (intensity)

The higher the

0 2
average B-factor,
the faster the
drop-off. How do we obtain
50 2 10 2
electron density from
This makes it more structure factor
1/res difficult (impossible) amplitudes and phases?
Low resolution High resolution to collect high
resolution data

The Fourier transform Fourier analysis

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

1/|qh| X-ray crystallography course 2006, Karsten Theis, UMass Amherst

|F|(h) Color: (h)
h=0
(r) |F|(h) Color: (h)
h=1
h=2
h=3
h=4
h=5
r h (r)
0 1 0 5
h
0 0 5
1
Real space Reciprocal space
Fourier analysis Real space Reciprocal space
Fourier synthesis
 X-ray crystallography course 2006, Karsten Theis, UMass Amherst X-ray crystallography course 2006, Karsten Theis, UMass Amherst X-ray crystallography course 2006, Karsten Theis, UMass Amherst

1
r

r
1

1
(r)

(r)

(r)
0

Fourier ripples
Fourier synthesis

Fourier synthesis

Fourier synthesis

0
(r)

Contour

2 sigma
level
add

add

add
h

h
5

5
|F|(h)

|F|(h)

|F|(h)
0

0
X-ray crystallography course 2006, Karsten Theis, UMass Amherst X-ray crystallography course 2006, Karsten Theis, UMass Amherst X-ray crystallography course 2006, Karsten Theis, UMass Amherst
r

r
1

1
(r)

(r)

(r)
0

0
Fourier synthesis

Fourier synthesis

Fourier synthesis
add

add

add
h

h
5

5
|F|(h)

|F|(h)

|F|(h)
0

0
 Building a model into the electron density
Interpreted electron density involves interpretation and prior knowledge

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

Protein/solvent regions
C-alpha trace
main chain, peptide direction
sequence assignment
side chain conformations
disulfides, metals, glycosylation and other
surprises

Questions: structure solution Assessing overall quality of structures

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

1) Which pairs of amino acids have very similar electron Quality criteria
density and are thus difficult to distinguish Resolution (related to # of observations per # of atoms)
crystallographically? Asp/Glu ; Thr/Val ; Leu/Ile ; Rfactor and free Rfactor (compares measured intensities to those
Lys/Met ; Asp/Asn ; Leu/Asp ; Glu/Gln calculated from the crystallographic model; <20% is best)
2) Which amino acids other than histidine have two side Geometry (Ramachandran plot, bond lengths and angles)
chain conformations resulting in almost identical electron Publication date (determines whether certain methods were
density? What could help to distinguish the two possible available, i.e. refinement, free Rfactor + other checks)
conformations?
Who checks the crystallographer?
H The reviewers
The protein data bank
N
N

N
N
Competing labs working on similar structures
H

Problematic regions within a structure

X-ray crystallography course 2006, Karsten Theis, UMass Amherst

1) Region is wrong in the model, i.e. different from true

conformation in the crystal (check electron density)
2) Region is absent from the crystallographic model
a. unstructured
b. folded but moves independently of rest of protein
3) Region was correctly modeled but
a. Conformation in solution is different
b. Region is flexible in solution but not in the crystal
c. Conformation changes upon ligand binding