Macromolecular Crystallography

1
Macromolecular Crystallography
Synchrotron Radiation Summer School Chester, Sep
2004
Elspeth Garman, Laboratory of Molecular
Biophysics, Oxford University.
elspeth_at_biop.ox.ac.uk
2
WHY? WHAT IS A MACROMOLECULE? OVERVIEW.
THE CRYSTALS. THE PROCESS. INTERPRETING THE
STRUCTURE
3
WHAT IS A PROTEIN? Made of amino acids. 20
different R groups in Nature.
R1
OH
H
C
N
C
CH
H
O-
H
R can be H, CH3, C2H5, C3H7, rings, CNH4N, CSH3
Basic units join to form chains.
4
20 amino acids in Naturemethionine and cysteine
contain sulphur
Red oxygen Blue nitrogen Yellow sulphur White
carbon Hydrogens not shown
5
e.g. Tumour Necrosis Factor
6
(No Transcript)
7
Ribbon monomer
TNF
Ribbon trimer
Space filled trimer
8
The structures of many proteins have now been
solved the first was Myoglobin (Kendrew et al
1960) which binds a heme (iron containing) for
function and structure. First enzyme solved was
lysozyme (Blake et al 1965)
9
From X-ray diffraction, we get experimental
electron density (green) and fit known sequence
of amino acids.
Alpha helix
DNA berinil
10
(No Transcript)
11
FMDV SRS 1988
12
100nm
Molecular structure determination by X-ray
crystallography
e.g. Structures of whole viruses (RNA) 4Å
resolution
13
Similar folds for the same function.
14
(No Transcript)
15
Substrate binding sites 2.0Å- 2.8Å
DRUG design
e.g. Neuraminidase from influenza virus (tern
N9) Relenza
16
STNA Iso-Carba-Dana at 0.91Å Seeing hydrogen
atoms like this is rare.
17
WHY? WHAT IS A MACROMOLECULE? OVERVIEW.
THE CRYSTALS. THE PROCESS. INTERPRETING THE
STRUCTURE YOUR QUESTIONS
18
(No Transcript)
19
Three dimensional structural information
complements other biochemical techniques. There
s a range of detail and information that can be
gained, depending on degree of order of
crystals.
20
The Unit Cell
b
The unit cell is a basic parallelopiped shaped
block from which the whole volume of the crystal
may be built by repetition in 3 dimensions. Any
point in the unit cell may be specified w.r.t.
the origin by parameters x, y, z measured
parallel to the unit cell axes and expressed as
fractions.
21
(No Transcript)
22
P
P,C
P,C,I,F
P,I
P
P
P,I,F
P
I
C
F
23
E/M wave
Frequency ? 1/period
Amplitude, A
2
A
Time
0
-2
A(t)A cos (2??t)
period
24
Amplitude
A(x)A cos (2?x/?) A cos (-2?x/?)
A
2
distance
0
-2
y(x,t) A cos 2?(?t-x/?)
?1Å T 3 x 10-19 s
25
DIFFRACTION from crystals If waves combine in
phase constructive interference. Bragg
satisfied n?2d sin ? If waves are out of phase,
they destructively interfere.
?
?
26
Blue and red waves have different PHASES
27
Combine as one diffraction spot
28
EVERY atom contributes to EVERY
diffraction spot. Conversely EVERY spot has a
contribution from EVERY atom.
Phase is determined by distance of atom
from reference plane (edge of unit cell).
29
DIFFRACTION PATTERN ? Spacing of spots depends
on size of unit cell. ? Intensity of spots
depends on structure of protein. RECIPROCAL
SPACE ? Big protein big unit cell, diffraction
spots close ? Small protein small unit cell,
diffraction spots far apart. i.e. spot
separation ? 1/unit cell
N.B. For crystals of 2 different proteins which
had identical unit cells, the spot spacing would
be IDENTICAL, the intensities of spots would be
DIFFERENT.
30
100K
31
Atomic resolution 1.2Å to 0.9Å N.B. VERY FEW
protein crystals give this resolution
32

• The wave scattered by a small volume dv at r
  relative to the wave scattered at the origin will
  have
• amplitude ?(r)dv
  and phase 2?r.S
• Hence the wave is described by
• ?(r) exp(2?ir.S) dv
• The total wave from the atom is thus
• f(S) ?vol of atom ?(r) exp(2?ir.S) dv
• Note f(0) ?vol of atom ?(r) dv
• Z (the number of electrons)

33
Structure factor equation

• The scattering by a unit cell with atoms at
  positions r1, r2, rj is given by the sum of the
  waves scattered by each of the atoms
• F(S) ?fj exp(2?irj.S) dv summed over the total
  no. of atoms, N.
• Amplitude of scattered wave is ? F(S) and the
  number of unit cells.
• The scattering by a crystal results in
  diffraction spots with indices hkl, hence get the
  Structure Factor equation
• F(hkl) ? fj exp(2?i(hxj kyjlzj)) summed over
  N atoms.
• F(hkl) F(hkl) expi?(hkl) where expi?(hkl)
  is the phase.
• I(hkl) F(hkl)2 expi?(hkl) exp-i?(hkl)
  F(hkl)2

34
Structure factor, electron density equations
35
Phases matter more this even applies to Nobel
prize winners !

• Jerome Karle (left)
• Herb Hauptman (right)
• FKarle, fHauptman
• FHauptman, fKarle

Methods in Enzymology 276 (1997) Randy Read
36
WHY? WHAT IS A MACROMOLECULE? OVERVIEW.
THE CRYSTALS. THE PROCESS. INTERPRETING THE
STRUCTURE YOUR QUESTIONS
37
(No Transcript)
38
CRYSTALS ordered arrays of molecules most
between 40 and 70 solvent
Size biggest ? 2mm x 1mm x 1mm smallest
currently useful ?30?m x 10?m x 5?m Hanging or
sitting drop crystallisation. N.B. New ROBOTS and
nano-drops (50nl-200nl).
Cover slip
Hanging drop containing protein in 25 sat AmSO4
grease
Water vapour
Resevoir of 50 sat AmSO4
39
Crystal growing is rather unscientific

• Need reasonable volume of protein.
• Molecular biology modification of protein may be
  necessary
• e.g. hepatitis A protease. 1 surface cysteine
• taken out.
• HIGH PURITY OF PROTEIN VITAL
• degree of homogeneity can pivotally affect
  chances of crystallisation.
• pH, temperature, organics, salts, ions,
  gravitygrandmothers maiden name etc.

40
(No Transcript)
41
WHY? WHAT DO I GET OUT? OVERVIEW. THE
CRYSTALS. THE PROCESS. INTERPRETING THE
STRUCTURE PAST QUESTIONS YOUR QUESTIONS
42
Phases, Fourier Transform
Diffraction Images
Electron Density model
43
Over expression/produce protein
Crystals
Initial diffraction F(protein) resolution
Derivatisation / Se-Met
Solve phases MIR/MAD/SAD
Initial structure
Iterative refinement
Characterisation/Quality
44
Basic Experimental
Optics
X-ray source
crystal
2?
Synchrotrons Lab sources
2.0? gt ?? gt 0.1?
goniometer
Detector CCD Image plate
Crystal gt 30?m synchrotron gt 100?m home
45
In-house X-ray Facilities

46
Most common commercial imaging plate detectors
Mar345, Mar300, Mar180 (MarResearch)
47
Lab source not bright enough for small or very
thin crystals. e.g. cyclin A, 5?m x 100?m x 300?m
48
Small crystals and large unit cells use a
synchrotron produced X-ray beam. Gives a very
parallel beam (small spots). Variable incident
X-ray ?. Very high flux small and weakly
diffracting crystals can be used.
Radiation emitted by accelerating electrons.
ESRF Grenoble
49
CCD detector schematic
Xray-to-light Convertor (Phosphor)Ga Arsenide
Eu
Fiberoptic faceplate (optional)
Demagnifyng Optics (Fibreoptic taper)
Light sensor (CCD)
X-ray (xph)
Electrons(e-) 20 lph ? 20e-
Light (lph)
40?m 500-700 lph/xph
Computer image
Readout electronics
Digital Units (ADU)
6-25 ?m pixel size (N.B.Optics)
still expensive
50
300?m beam
Room temperature HEWL crystal after 3 hours in
a 2nd generation synchrotron beam. In high flux
beams, cryo-temperature data collection (100K)
is essential, as it reduces radiation
damage. Antifreeze (cryoprotectant) added.
51
Haas and Rossmann 1970 lactate
dehydrogenase Acta Cryst B26, 998-1004.
52
significantly reduced at 100K
PRIMARY inevitable, a fact of physics! SECONDARY,
can we control it?
Proportions?
53
293K
100K
54
Other advantages

• Usually get a whole data set from a single
  crystal ? higher QUALITY data.
• For MAD, the systematic errors are minimised
  by using only one crystal.
• Can harvest and store crystals while they are
  in peak condition.
• Small crystals and flat plates can be mounted
  easily.
• No secondary radiation damage during storage.
• New experiments are possible.

55
(No Transcript)
56
Cryo-buffer optimisation replace the water
in mother liquor with cryo-agent, rather than
diluting it. test cryo-buffer in loop alone.
57
Cryo-buffers.

• PEG lt 4K? increase PEG, add small PEGs
• PEG ? 4K? add small PEGs
• 2/3rds of cases ? add 15 - 25 glycerol
• MPD ? harvesting buffer, increase MPD
  concentration.
• Salt ? add MPD and/or ethylene glycol or glycerol
  
• ? increase conc/add salt
• ? Exchange salt. e.g. 100 8M Na Formate.
• N.B. Low salt needs gt concentration of
  cryoprotectant than high salt.
• Sugars, oils (paratone N), combinations,

Cryocrystallography See Garman and Schneider,
JAPC, 1997
58

• Fishing
• minimise handling
• minimise liquid round crystal
• fish near cryogen
• loop perpendicular to liquid
• N.B. Acupuncture needles.
• Salt crystals in loop.

59
Synchrotrons even at 100K there is a problem.
Bacteriorhodopsin 10?m thick crystals, 30?m
diameter microbeam.
Crystals allowed to warm up after 100K data
collection
60
Centre crystal in the beam
Beam must hit intersection of axes. Crystal at
intersection. X wires may not be.
?
detector
beam
?
2?
TV X wires
Where is the beam? Home check with pin.
Synchrotron pink/green paper.
61
Image from E. Garman
62
Crystal diffraction

• Salt? Do a large ?? image.
• Obviously twinned
• Internally twinned
• Disordered high mosaic spread, disordered along
  one axis, statistical disorder.
• Diffraction weak.
• None
• If good, what is resolution limit? Reassess
  crystal to detector distance.
• Reasonable mosaic spread
• Spots are resolved
• Spots are not overloaded

63
e.g Twinned crystal with 2 distinct lattices.
64
Phi 0
Phi 90
Always check diffraction in two orthogonal images
!
65
Spots have to be separated.
66
What ?? ?2D data collections
?? gt M
Ewald sphere
Oscillation end
Oscillation start
Fully recorded reflection
Counts in spot
1 2 3 4 5
x
Image number
67
What ?? ? 2D data collections
?? gt M
Ewald sphere
Image 2
Image 3
Oscillation start
Partially recorded reflection
Counts in spot
1 2 3 4 5
Image number
68
What ?? ? 3D data collectionsfine ? slicing
?? lt M
Ewald sphere
4
Oscillation start
3
2
5
1
3D profile fitting of reflection possible
DETECTOR
Counts in spot
x
y
y
y
Image 2
Image 3
Image 4
1 2 3 4 5
x
x
x
Image number
69
From Dauter, Methods in Enzymology (1997), 276,
P326 -344
70
(No Transcript)
71
(No Transcript)
72
Overlaps

• ??max ? dmax - ? radians
  ? 180/?
• a
  to get ?
• N.B. Mosaicity, ?
• e.g. 1.5Å data, max cell 128Å
• 1.5 - ? 0.012 - ? radians 0.67?- ?
• 128

73
Why mosaic spread affects maximum
oscillation angle.
Fruit!!
74
Total rotation required for complete data in case
of symmetric detector position

• Rotation required for
• Crystal class Point group Standard
  data Anomalous data
• Triclinic 1 180? 360?
• Monoclinic 2 180? (b?), 90? (a?, c?) 180? (a?,
  b?, c?)
• Orthorhombic 22 90? (a?, b?, c?) 90? (a?, b?,
  c?)
• Tetragonal 4 90? (a?, b?, c?) 90? (c?), 180?
  (a?, b?)
• 422 45? (c?), 90? (a?, b?) 45? (c?), 90? (a?,
  b?)
• Trigonal 3 60? (c?), 90? (a?, b?) 60? (c?),
  90? (a?, b?)
• 321 30? (c?), 90? (a?, b?)
• 312 30? (c?), 90? (a?, b?)
• Heaxagonal 6 60? (c?), 90? (a?, b?)
• 622 30? (c?), 90? (a?, b?)
• Cubic 23 About 60? About 70?
• 432 About 35? About 45?

75
(No Transcript)
76
Collect complete data low resolution data
are important. 838 reflections missing from
hole
With hole main chain electron density
uninterpretable Without hole electron density of
protein ligand easily interpretable
77
Resolution dmax (Å)
2? 0
detector face
2?max
S
beam
L
2?max
crystal
Bragg n? 2d sin ? dmax ?
2 sin ?max
tan 2?max L 2S
e.g. L 345mm, S 184mm, ?1.54Å gives dmax
2.1Å
78
Resolution of pattern (how far out the spots
go) is directly related to amount of structural
information gained. From data get (h, k, l) and
I, error on I for each spot. Find phases
MAD/MIR/Molecular replacement N.B. addition of a
heavy atom will change the intensity of EVERY
spot in the pattern. From sqrt (I) (F) and
phases, can calculate an electron density
map. Known primary sequence then built into
density automatically or by hand. This is the
starting model.
79

• Multi-pass data collections
• e.g. for a Mar345 IP
• Resolution (Å) ?? S(mm)
• 0.83 2.0 0.5? 110
• b) 1.35 8.0 0.8? 250
• c) 2.66 25 1.5? 550
• shadows at UHR, overlaps

80
Resolution of the map
5Å See the shape of the molecule, and helices
stand out as rods At 4Å, even b-sheets
traceable 3Å Usually get chain fold right, and
see where side chains are. Since 2.5 Å, should
have seen the main chain decorated with carbonyl
bulges as well as clear choice of side chain
rotamers. 1.5 Å essentially atomic resolution
most atoms in the right place.
81
Backstop position important if cup is near
crystal, main beam air scatter is much reduced.
82
Backstop distance have to compromise!
Asymmetric backstop poor low resolution Rmerge.
83
The Phase Problem
84
The phase problem

• Molecular Replacement
• Multiple Isomorphous Replacement (MIR)
• Multi or single wavelength anomalous dispersion.
  (MAD/SAD)
• Most proteins too big to use direct methods
• (biggest solved this way to date is 1000 atoms)

85
(No Transcript)
86

Green STNA Red Flu N9
Too dissimilar for MR (15 sequence homology)
87
MIR Multi-isomorphous replacement
Difference map between native and derivative
shows Xenon atom bound.
88
Oxford Cryosystems Xcell.
89
MAD multi-wavelength anomalous dispersion.
Maximise the Anomalous (red) or Dispersive
(green) Difference? If dispersive, then collect
inflection point last to maximise
dispersive Difference from radiation damage and
f If Anomalous, remember that collection above
the edge damages much more rapidly than below it.
90
WHY? WHAT IS A MACROMOLECULE? OVERVIEW.
THE CRYSTALS. THE PROCESS. INTERPRETING THE
STRUCTURE
91
Quality of structure

• Resolution- detail
• R-factor from refinement of model to data
• Ramachandran, ?-? plot, (? trans/cis)
• Bond lengths and angles
• B-factors temperature factors
• Pseudo Energy minimised (bond lengths and angles
  crystallographic term)
• Does it make sense re other biochemical data?

92
3D structure detailed biological information.