Macromolecular

Crystallography

Jeroen R. Mesters
University of Lübeck
 3-D structure methods
Method Principle Needs Resolution Molecular Examples
Weight
X-ray diffraction 3D crystals 0.1 nm 107 protein
from virus
electrons ribosome
NMR spin concentrated ? 5·104 protein
transition solution (inc. compounds
labels)
Neutron diffraction big 3D 0.2 nm 107 protein
diffraction from nuclei crystals
Electron diffraction 2D crystals 0.3 nm 107 membrane-
Diffraction from protein
electrons
Electron light intact 1.5 nm „no limit“ cell-surface
microscopy microscope particle ribosome

Diffraction power 1:100:1000=electron:X-ray:neutron=thin→big sample

 Why X-rays?
• Ernst Abbe (1840-1905), german mathematician and physicist, professor at
Jena University, worked with Carl Zeiss. Developed modern microscope
theory.

• Abbe's law of limiting resolution: d = λ / (2·NA)

(d: spacing, λ: wavelength and NA: limiting numerical aperture of objective,
1 if objective lens is large enough to capture the first-order diffraction):

the minimal resolvable distance between two points depends on the

wavelength of the light used, Dmin ≅ λ/2 (for visible light 4-8·10-7 m, Dmin ≅
2·10-7 m)
C – C distance about 1.5 Ǻ

Röntgen
~105 eV → 0.4–2.5 Ǻ (λ = 12398/E [Ǻ])
1Ǻ = 0.1 nm = 10-10 m
Abbe
 Why crystals?
http://www-structure.llnl.gov/Xray/101index.html

One molecule More ordered molecules

Very weak diffraction More discrete diffraction

 Historical hallmarks
• The diffraction of X-rays by crystals was first demonstrated by Friedrich,
Knipping and von Laue (17 years after the discovery of X-rays by Röntgen
in 1895): unknown at the time were the nature of both X-rays (particles or
electromagnetic radiation) and crystals (internal periodicity or not).

• The first small molecule structures of KCl, NaCl, KBr and KI in 1913
(Bragg).

• First protein crystal (hemoglobin) in 1830, first recorded protein diffraction

in 1934, DNA double helix structure in 1953 (Watson & Crick) and first
protein structures of myoglobin (Kendrew) and hemoglobin (Perutz) in
1959.

• Nobel Prices
Röntgen 1901; Laue 1914; W. Bragg 1915; Watson & Crick 1962;
Kendrew & Perutz 1962; Deisenhofer Huber & Michel 1988.
 Crystal growing: all about
tampering with the solubility
of the protein

(“cristallos“ = clear ice)

 Solution Properties of Proteins
Physicochemical properties:
- surface properties → pI, post-translational modifications
pH:
- pH extremes – fold disruption
- if pH = pI → solubility↓
Temperature:
- class I – ↑ solubility with ↑ temp., most common
- class II – little or no temperature effect
- class III – ↓ solubility with ↑ temp.
Miscellaneous:
- buffers, salts, detergents, organic compounds
- ligands, cofactors
 Protein/Salt mixtures
The Hofmeister Series (1888)

Precipitating Power:
Anions: sulphate > phosphate > acetate > citrate > chloride >
nitrate >> chlorate > thiocyanate

Cations: Li+ > Na+ > K+ > NH4+ > Mg2+

Most stabilizing > Most destabilizing

Salting out > Salting in

- Negative net charge collagenese (pI 4.1, set up pH 7.2)

phosphate > sulfate > citrate > chloride
(ammonium sulfate with some sodium chloride)

- Positive net charge lysozyme (pI 9.5, set up pH 4.8)

thiocyanate > nitrate > chloride > acatate > citrate
(inversion of Hofmeister series!!)
 Main Salt Classes

Kosmotropes – strong H2O interactions

water-structure makers
singly- or multiply-charged ions
high charge density
SO42-, HPO42-, Mg2+, Ca2+, Li+, Na+, H+,
OH-

Chaotropes – weak H2O interactions

water-structure breakers
large, singly-charged ions
low charge density
SCN-, H2PO4-, HCO3-, I-, Cl-, NO3-, Chaotrope
NH4+, Cs+, K+, (NH2)3C+ (guanidinium)
 Kosmo- and Chaotropes
- 1st hydration shell is ordered and dense
Prot Prot (~20% denser than bulk water)
- regions of high water density (low
entropy) termed 'icebergs‘.
- icebergs help maintain folded state
1st hydration Bulk
- icebergs keep protein molecules separated
shell (dense) water
from each other

Prot Prot Prot Prot

Kosmotropic effect Chaotropic effect

• ordering the bulk water • disordering hydration shell
(melting the icebergs)
 Solubility diagram f([salt])
Protein solubility

Salting in Salting out

Inverse Salting in

[salt]
 Solubility diagram f(pH)
Protein solubility

← Salts - PEG – Salts →

Crystallization slot (!?)

1-2 pH units away from pI

pI

pH
 Phase Diagram

growth
[protein]

nucleation

precipitation

labile zone
soluble metastable zone

[precipitant]
 Concentrated Protein Solutions
[protein]

For NMR and Xtallography alike:

In order to prepare a highly

concentrated solution that keeps the
protein happy in solution, one needs to
screen (!) for the best starting
soluble conditions with respect of the right
buffer, pH and salt etc. There is no
[precipitant] thing like „one buffer fits all“.

Trouble already starts at the level of cell disruption!

) sparse matrix screening
 Crystallization methods

• Vapor diffusion

• Dialysis

• Batch

• Capillaries
 Protein purity/concentration
Naturally, both as high as possible but,
sufficient amount left to screen with…….

If concentration via centrifugation or dialysis

(„protein sticks to membrane“) does not work,
immediately check for aggregation!
 Precipitants
Salts: (NH4)2SO4 NaCl K/Na-phophates Malonate/Tartrate

Polymers: Poly-ethelene-glycol (different MW: 400 -20.000)

Organic solvents: 2-Methyl-2,4-pentanediol Ethanol Acetone

Buffers: Acetate Citrate Tris Hepes Cacodylate

Detergents: ß-octylglycoside Triton-X100 LDAO

Additives: EDTA Spermin MgCl2 ATP CaCl2 Zn-acetate

Ligands Inhibitors etc.
Vapor diffusion
 Phase diagram
growth
phase separation
[protein]
(oil droplets)
precipitation

nucleation

spherulites
labile zone
metastable zone
soluble

[precipitant]
 Vapor modification

Oil layer: Al‘s oil or silicon oil

Speed of equilibration:
-Salts fast (~2 days)
-Polymers slow (4-7 days)

Oils work best for salts

Courtesy: JuanMa García-Ruiz, http://lec.ugr.es

 Protein with or without sugars?
• Glycosylation leads to heterogeneities
• Deglycosylation (or expression in E. coli)
often leads to inactivation

94
67

43

30

20

14,4
1 2 3 4 5

12,5% SDS-PAGE Gel of IEF-PAGE gel (pH 3-9) of protein

Garlic the protein stock solution. stock solution and different crystals.

E. Bartholomeus Kuettner, Rolf Hilgenfeld and Manfred Weiss, J. Biol Chem. 277, 46402-46407.
 H9

N328-A N328-B N276

K280*
H6

H11 H11
Y174

N146-A R169*
H5 H5
K173*
S10 S10

A149 T148
N146-B N146-A

Space group P212121 Resolution range 20 - 1.53 Å

R factor of 19.3% (free 22.9%)

Sugars (N146) involved in monomer-monomer contacts within the

homodimer.

E. Bartholomeus Kuettner, Rolf Hilgenfeld and Manfred Weiss, J. Biol Chem. 277, 46402-46407.
One more example
PAP-S, isolated from the seeds
of pokeweed, belongs to the
family of the type-1 ribosome
inactivating proteins (site
specific depurination of the α-
sarcin/ricin loop in rRNA).

Crystals of PAP-S were grown

from a heterogeneous mixture
of two isozymes, approximately
29 and 30 kDa.

pokeweed
Tanis Hogg et al., Acta Cryst. D58, 1734-1739.
Space group I222 No less than two direct and two
Resolution range 30 - 1.7 Å water-mediated hydrogen bonds
R factor of 18.1% (free 21.9%) are formed.

Tanis Hogg et al., Acta Cryst. D58, 1734-1739.

Microdialysis(-button/-rod)

Easy to adjust/correct!
Good idea to start with….
 Phase diagram

growth
[protein]

nucleation

precipitation

labile zone
metastable zone
soluble

[precipitant]
Remember, easy to adjust/correct!
Good idea to start with….
(Micro-)batch under oil
 Phase diagram
growth
phase separation
[protein]
(oil droplets)
precipitation

nucleation

spherulites
labile zone
metastable zone
soluble

[precipitant]
Only true for evaporating dropslets (silicon or Al‘s oil)
Membrane proteins under oil

Oil has an advantage

….slowly absorbing the
detergent thereby
promoting crystal
growth….
 Capillary

gel
Courtesy: JuanMa García-Ruiz, http://lec.ugr.es

Nail Polish
Precipitant + glycerol + heavy metals Nail Polish

Protein and low concentration agarose

 Crystal properties
• Mechanical, electrical and optical properties, symmetry, etc.

• In general, a crystal is an anisotropic body with anisotropic properties like

ist dielectrical constant, ist polarization and ist refraction index:

Optical activity:
triclinic, monoclinic and orthorhombic crystals polarize light along all axis
trigonal, hexagonal and tertagonal crystals are isotrope // C
cubic crystals are isotrope

• Due to the high water content (30-80%), protein crystals can be soaked
with heavy metal compounds and dyes, but also substrates (ribosome
crystals with tRNA).

• A protein crystal is soft whereas a salt crystal is „rock hard“ (needle test).
 Salt or protein

- Birefringence
- IZIT
- Needle test

- Ultimately, X-ray
diffraction
 Crystal gallery

Beauty is only skin deep, appearances can be deceiving (big crystal, no diffraction)
Micro- and macro-seeding for growing bigger crystals (cat whiskers)
Epitactical growth of protein crystals on an amorphous surface
 X-ray sources

• Sealed tube

• Rotating anode

• Synchrotron
 Sealed tube
Cu
anode
Ni-filter

X-rays
e-

M level
cathode

L level
Kα1 1.54051 Å
1.54178 Å
Kα2 1.54433 Å Kα2 Kα1 Kß
K level
Kß 1.39217 Å Ni-filter
 Synchrotron
X-rays

e-

Very intense light with tunable wavelenght

http://unisgi2.desy.de/x13.html
Diffraction Pattern
 Interference of waves

+ A
+
λ

= =
2A

„completely out of phase“

„in phase“ (constructive) destructive interference
path difference = n·λ path difference = (n+½)·λ
 Bragg‘s law
Path difference between waves 1 and 2 is
1 equal to OA + OB. In the case of
constructive interference, OA + OB = n·λ
θ
2

θ
Since θ
d OA = OB = d · sinθ
A B it follows that (OA + OB =)
θ
n·λ = 2d · sinθ
O Bragg‘s law
 Number of reflections
How many reflections are possible at a given λ is equivalent to asking how
many lattice points lie within the limiting sphere

Volume of limiting sphere is 4/3 · π · r3 = 4/3 · π · (2/λ)3

(because r of limiting sphere is 2·r of reflection sphere)

Number of reflections N = 4/3 · π · (2/λ)3 · 1/(reciprocal cell volume)

N = 33.51 · (volume direct cell) / λ3

Example: P222 (1 r.l.p. per U.C.) with a = 150 Å, b=100 Å and c=40 Å

λ= 1.54 Å ) N = 33.51 · 600000 / 3.6523 = 5.505.024 reflections

λ= 0.80 Å ) N = 33.51 · 600000 / 0.512 = 39.269.531 reflections

Luckily, these are not all unique reflections and do not all need to be measured!
 Generalised structure factor
Fourier pair of equations:

F(hkl) = V ∫x∫y∫z ρ(xyz) exp[2πi·(hx + ky + lz)]dxdydz

ρ(xyz) = 1/V ∑h∑k∑l |F(hkl)| exp[-2πi·(hx + ky + lz) + iα(hkl)]

y y
Discrete Continuous

0 1 x x
f(x0→1) = x1 + x2 +x3 + x4 + x5 f(x0→1) = ∫0→1 f(x)dx
= ∑x f(xn) {the area under the function curve}
 Information loss

From our diffraction pattern we can determine the relative values of F(hkl)
because they are proportional to I(hkl). However, we can not determine
α(hkl)! Bad news, we can not calculate the structure immediately.

ρ(xyz) → I(hkl) → |F(hkl)|2 {no α(hkl)}

↑

The so-called „crystallographic phase problem“ can be solved by using some

cleaver methodologies: MR, MIR, MAD, etc.

Ihkl ≡ Io· λ3/ω · Vx · L · p · A/V2 · |Fhkl|2 (C.G. Darwin)

 What can be calculated?
• We can still calculate a Fourier summation with the
intensities as coefficients and all αhkl equal to zero,
e.g. a Patterson map.

• P(uvw) = 1/V·∑h∑k∑l |Fhkl|2·exp[-2πi·(hx+ky+lz)]

= 1/V·∑h∑k∑l |Fhkl|2·cos[2π·(hx+ky+lz)]

≡ ∫v ρ(x,y,z) ρ(x+u,y+v,z+w) dV

* Note that eix = cos(x) + isin(x).

 ∫v ρ(x,y,z)ρ(x+u,y+v,z+w)dV

Ori

(0.2 0.2)

ρ(xyz) P(uvw)
(0.1 0.5) (0.5 0.5)
 Consequences
• Only interatomic vectors in real space show up as peaks in Patterson
space since P(uvw) = ∫v ρ(x,y,z) ρ(x+u,y+v,z+w) dV.

• The map of a real unit cell with N atoms will contain N2-N peaks in
Patterson space outside the origen. Origin will contain N peaks.

• Patterson maps contain an additional symmetry element

(centrosymmetry).

• In simple structures with a limited number of atoms, the atomic

positions can be derived fairly straightforward.
MIR Fph = Fp + Fh

Cosine rule c2 = a2 + b2 -2ab·cos(γ)

cos(γ) = - (c2-a2-b2)/2·a·b

γ = 180º + αp - αh
Fph Fh
π−αh αh
αp
Fp

αp αph
 Harker construction
|Fp| |Fph2|

cos(180º + αp – αh) = - cos (αp - αh) =

- (|Fph |2 - |Fp |2 – |Fh |2) / 2·|Fp|·|Fh|
|Fph1|
αp= αh + arc cos(|Fph|2-|Fp|2-|Fh|2)/2(|Fp||Fh|)

Two solutions for one derivative.

More derivatives will solve the ambiguous result!

 Multi-wavelenght Anomalous Dispersion

• Normal situation, |Fhkl| = |Fhkl| with αhkl = - αhkl

• The most tightly bound electrons in atoms (atom specific,

wavelength dependent) cause anomalous scattering of X-rays.
From Sulfur onwards, a measurable effect occcurs that causes
a difference in intensity in the Bijvoet pairs, |F+|2 and |F-|2.

• For anomalously scattering atoms,

|Fhkl| ≠ |Fhkl| with αhkl ≠ - αhkl

• A synchrotron beamline with tunable wavelenght is needed.

Metals in crystallography

Zn
λ=1.28Å
9661 eV
CuKα

X-ray anomalous scattering

 HK0
layer

X-ray anomalous scattering hkl ≠ hkl

 About f‘ and f‘‘
eV f‘ f‘‘
low large small

inflection minimum large

peak medium maximum

high large medium

In principle, a 2-wavelength experiment

Peak f‘‘ maximal
should solve the phase problem. Inflection point |f‘| maximal
 Finale

May the crystal force be with you…..

Thanks for your attention!

Rotspon
Marzipan