Organic and macromolecular crystallography with electrons€¦ · Tim Grüne Electron Crystallography nanoArgovia A3EDPI Spot Intensity • Spots intensity depends on physics of interaction

Tim Grüne Electron Crystallography nanoArgovia A3EDPI

Organic and macromolecular crystallography with electrons

Bioinformatics and X–Ray Structural Analysis — Universität Konstanz29th January 2019

Dr. Tim Grünedes. Head of the Centre for X-ray Structure AnalysisFaculty of ChemistryUniversity of Vienna

Universität Konstanz 29th January 2019 1/0


1 - Outline

1. Structure Determination with Crystallography

2. Electron and X–rays

3. Applications for Electron Crystallography

4. Practical Aspects

5. Radiation Damage

6. Dynamic Scattering



2 - Structure Determination by Single Crystal Diffraction



The Crystallographic Experiment

Detector

Crystal

SoRadiation

source

Planar

wave

S i

Rotation Axis of Crystal

(Diffr

action)

2θ

Data Noise (ignored)

• Diffraction spots: interaction between wave and crystal

• Experimental result: Position and Intensity for each spot

• Data recorded as frames with an area detector

• Generally true for all radiation sources, X-ray, neutron, electron



Spot Position

• Spots positions according to Laue Conditions and orientation of Unit Cell:

(~So−~Si).~a = h

and (~So−~Si).~b = k

and (~So−~Si).~c = l

• Monochhromatic wave: ~S = (~So−~Si) depends on wavelength λ and experimental geometry

• Spot position ⇔ Crystal lattice, independent from radiation type

• Resolution dhkl of a spot from position on detector via Bragg’s law, λ = 2dhkl sin(θ)



Spot Intensity

• Spots intensity depends on physics of interaction

X–rays interact with electrons, crystallographic map corresponds to electron density (number of electronper Volum, e−/A3).

Electrons interact with electrostatic potential from electrons + nucleus (ϕ(~r))

Neutrons interact with nucleus via weak interaction, and magnetic moment.

• Spot intensity ⇔ Unit cell content: where are the atoms, what type of atoms are they



Types of Radiation — X–rays

Annual Growth of the PDB (X–ray)

• most advanced (pipelines from data collection to structure refinement)

• typical wavelength: λ =0.8–1.9Å

• standard structure determination

PDB CSD(Macromolecules) (Organic Compounds)

X-ray 132,000 950,000neutron 152 1,500electron 111 ?



Types of Radiation — neutrons

PDB ID 2ZOI: D/H exchange in β–strand

(Gruene et al, J. Appl. Cryst. 47 (2014), 462–466)

1. visualisation of hydrogen atoms

2. adjacent elements (e.g. K+ vs. Cl−, Zn2+ vs. Cu+)

3. requires large crystals (≥ 1mm3)

4. (virtually) no radiation damage

5. structure determination from radiation sensitive samples (Photosystem II)



Types of Radiation — electrons

1. strong interaction compared with X–rays: good for very smallcrystals (≪ 1µm thickness)

2. typical wavelength: 200keV = 0.0251Å

3. electron optics: imaging and diffraction with the same instru-ment

Diffraction of ZSM5 zeolite catalysts

Gruene et al., ChemEurJ (2017)

DOI: 10.1002/chem.201704213



3 - Electron Diffraction



Small Crystals

Main advantage for electron crystallography:Diffraction from very small crystal (< 1µm)

Some instruments provide 5–10 nm beam diameter



Crystallisation of Proteins

Luft, Wolfley, Snell, Crystal Growth& Design (2011), 11, 651–663



Drops viewed through TEM

Stevenson,. . . , Calero, PNAS (2014) 111, 8470–8475 / Calero, . . . , Snell, Acta Cryst (2014) F70, 993–1008



Electron Diffraction: Nanocrystals

organic compound

sucrose (ETH coffee bar)

Silicalite–1 / ZSM–5 (Teng Li)Gruene et al., Chem. EurJ (2018), 24, 2384–2388



How small is “nano”?

typical protein crystal size for X-rays

typical protein crystal size for elec-trons, 100x140x1,700 nm3

volumes compare like 1m3 or 6 bath tubs of water vs. 10µl



Effects of Crystal Volume on Diffraction Data

Reducing crystal volume reduces the resolution by (at least) two effects:

1. I(hkl) ∝ Vcrystal: 1/10 volume = 1/10 intensity

2. Henderson / Garman limit: maximum dose per volume before resolution is halved: 1/10 volume = 1/10 dosebefore radiation damage destroys crystal

From (1): In order to record the same quality diffraction pattern from a 10 times smaller crystal requires 10 timesmore intense beam.

From (1)+(2): This makes the crystal die 100 times faster



4 - Instruments for Electron Diffraction



Electron Microscopes

(Wikipedia)



The Lens System

see e.g. Zuo & Spence, “Advanced Transmission Electron Microscopy”, SpringerCarter & Williams, “Transmission Electron Microscopy”, Springer

C3

C2

C1

Gun / Source

Objective lens

Sample

Diffraction Plane

Imaging Plane

• Lenses C1–C3 shape beam• Crystallography: Parallel beam• Objective lens: sets effective detector distance to

backfocal plane = diffraction mode• C3 not present in all microscopes

Lenses cause distortions.



Electron Microscope: Imaging Mode

Plane Wave Object Lense Image Plane (Detector)

Rays of equal origin focus on detector



Electron Microscope: Imaging Mode

Plane Wave Object Lense Image Plane (Detector)

Detector noise and ra-diation senstivity re-quire low contrast im-ages

Martinez-Rucobo et al.Molecular Cell (2015) 58,1079–1089



Electron Microscope: Diffraction Mode

Plane Wave Object Lense



Electron Microscope: Diffraction Mode

Plane Wave Object Lense

Backfocal Plane

Rays of equal direction focus on detector

“Image” corresponds to Fourier transform of object

Image at Backfocal Plane =‖Fouriertransform of object‖

If object = crystal:

diffraction spots according to Laue conditions



Detectors for X-rays suitable for electrons

• Eiger X 1M designed for X-ray Synchrotron radiation• 1030x1065 pixel, 75×75 µm2

• up to ν = 3kHz frame rate: data collection at synchro-tron speed

• 3µs dead time: shutterless data collection• ≤ 200keV : no radiation damage, no beam stop• 16 or 32 bit image depth & 2.8 · 10

6 ctss·pixel: high dy-

namic range

Diffraction pattern and structure for SAPO–34Tinti et al., “Electron crystallography with the EIGER de-tector”, IUCrJ (2018). 5, 190–199



Installation of the EIGER X 1M in 1/2 day (C. Zaubitzer, ScopeM, ETH)

Removal of the previouscamera

Mounting of the EIGER X1M with adapter flange

Shielding and radi-ation monitoring

• Final shielding after 1/2 day• Vacuum OK: next morning

• Return to original state: 1 day• Gatan camera back with auto-justage



5 - Data Collection: Experimental Parameters



The Geometry Section in XDS Input File

EIGER detector not connected to the electron microscope: experimental are not present in file header.

Wavelength λ ∝ 1/E; 200keV =̂0.02508Å

✦❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂ ●❊❖▼❊❚❘■❈❆▲ P❆❘❆▼❊❚❊❘❙ ❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂❂

❖❘●❳❂ ✺✶✺ ❖❘●❨❂ ✻✷✷ ✦❉❡t❡❝t♦r ♦r✐❣✐♥ ✭♣✐①❡❧s✮✳ ❖❘●❳❂◆❳✴✷❀ ❖❘●❨❂◆❨✴✷

❉❊❚❊❈❚❖❘❴❉■❙❚❆◆❈❊❂ ✹✾✺ ✦✭♠♠✮

❘❖❚❆❚■❖◆❴❆❳■❙❂ ✲✵✳✼✸✽✺✹✺ ✲✵✳✻✼✹✶✾✺ ✵✳✵✵✸✺✻✵

❖❙❈■▲▲❆❚■❖◆❴❘❆◆●❊❂✵✳✵✷✾✺✾✸ ✦❞❡❣r❡❡s ✭❃✵✮

❳✲❘❆❨❴❲❆❱❊▲❊◆●❚❍❂✵✳✵✷✺✵✽ ✦❆♥❣str♦❡♠

■◆❈■❉❊◆❚❴❇❊❆▼❴❉■❘❊❈❚■❖◆❂✵✳✵ ✵✳✵ ✶✳✵



❖❘●❳✴❖❘●❨: Beam Centre

• Strong interaction of e− with crystal ⇒ direct beamweak compared with X-rays

• Direct beam visible on detector• Coordinates manually from any image viewer, e.g.

ADXV (Andy Arvai,✇✇✇✳s❝r✐♣♣s✳❡❞✉✴t❛✐♥❡r✴❛r✈❛✐✴❛❞①✈✳❤t♠❧)

• on–axis detector: direct beam ≈ ORGX/ORGY• Can vary per data set



Detector Distance

• Detector distance set through electron optics• Detector physically not moved

• Calibration once per lens settings• Set manually with adxv and (Aluminum) powder–

pattern• Estimated accuracy: ±5mm



The Rotation Axis

• X-rays: rotation axis obvious from instrument setup• Electron Microscope: Objective lens determines (ef-

fective) detector distance• Image rotates step–wise depending on objective lens• powder pattern with sample oscillation (“α–wobbler):

spot on axis very strong, non–spot stays non–spot

Summed images for D560mm

Summed images for D380mm



Determination of the Rotation Axis

• Rotation axis runs through direct beam and minimumof powder ring

• line through 2 points on rotation axis• P1 = 529/621 and P2 = 458/702

• tan(α) = ∆Y∆X

= 702−621

458−529

• ❘❖❚❆❚■❖◆ ❆❳■❙❂ cos(α); sin(α); 0

• Large radius of convergence with XDS (≈ 5−10◦)

Direction of rotation: from minimal error of spindle axis

❘❖❚❆❚■❖◆ ❆❳■❙❂ (XDS.INP) -0.6979 -0.7161 -0.0102 +0.6979 +0.7161 +0.0102❉❊❱◆ ❖❋ ❙P■◆❉▲❊ P❖❙◆ (IDXREF.LP) 0.37

◦1.01

◦



Oscillation Width

• Current setup with “alien” detector: Sys-tem parameters not synchronized withdetector

• Step froward at C–CINA: movie duringmeasurements

dφ

dt=

∆α

∆t=

111.6◦−53.3◦

26.449s−1.264s= 2.315

◦/s

ν(EIGER) = 100Hz

⇒ ∆φ = 0.0231◦/frame



Oscillation Width

-80-60-40-20

0 20 40 60 80

-10000 0 10000 20000 30000 40000 50000

α [d

egre

e]

t [ms]

dφ/dt [°/s]= 2.9520 ± 0.0079, α0 [°] = -62.31 ± 0.19

datafile usi (0.5*($1+$3)-t1):2

α(t) = φt + α0

• Probe α angle per 0.5s during experiment• Fit line to measurements• fast, reproducible• Oscillation width ∆φ [◦/frame] = dφ

dt/ν(EIGER)

Acknowledged: Luca Piazza. Dectris Ltd. for initial Digital Micrograph script



6 - Dynamic Scattering



Dynamic Scattering

• Kinematic Theory of Diffraction: Every photon / electron / neutron scatters once in the crystal

• |Fideal(hkl)| ∝√

Iexp(hkl)

• Dynamic Scattering: Multiple Scattering events occur

• Electron Diffraction: Multiple Scattering occurs even with nanocrystals



Dynamic Scattering



Multiple (Dual) Scattering

~Si

~S1o

~S′o ~S2o

• First reflection ~S1o acts as source of

second reflection ~S′o.• Second reflection ~S′o overlaps with an-

other reflection ~S2o



Multiple (Dual) Scattering

~Si

~S1o

~S′o ~S2o

Laue Conditions (accordingly~b and~c):

(~S1o−~Si) ·~a = h1

(~S′o−~S1o) ·~a = h

′

(~S′o−~Si) ·~a = h1+h′ = h2

~S′o −~Si fulfills the Laue conditions, hence thesecondary reflection ~S′o−~S1

o overlaps with the“ordinary” reflection ~S′o−~Si.



Experimental Treatment of Dynamic Scattering for Organic Crystals

• Exact calculation very complicated

• Not feasible for complex molecules (more than a few atoms)

• Ignorance of dynamic scattering, i.e. assumption of kinematic scattering (as in X-rays) provides reliablestructures

• Data statistics and model statistics poor, despite reliable structures



7 - Example Structures



Grippostad R©, STADA

active compounds non-active compounds

paracetamol gelatineascorbic acid glycerol tristearate

caffeine lactose monohydratechlorphenamine maleate quinoline yellow (E104)

erythrosine (E127)titanium dioxide (E171)



Single Crystal Structure from a Pharmacy Powder

1. Exp: a = 6.9, b = 9.4, c = 11.6, α = 90.6, β = 98.4, γ = 89.8

2. CSD : a = 7.1b = 9.3c = 11.7α = 90.0β = 97.7γ = 90.0; CSD search: HXACAN04, P21/n, Paracetamol,

3. Structure solved with ≤ 40% completeness

4. Difference map reveals hydrogen atoms: data sensitivity

“The existence of multiple crystal forms (polymorphs, solvates, hydrates, etc.) is playing an increas-ingly important role in establishing and protecting intellectual property rights in the pharmaceuticalindustry”

(Prof. J. Bernstein, ECM-30 (Aug. 2016), MS50-O2)



8 - Structure of a New Methylene Blue Derivative MBBF4

I. Regeni, Dr. J. Holstein & Prof. G. Clever, TU Dortmund



MBBF4 — the Crystal (J. Holstein, TU Dortmund)

C120H120B8F32N28S4Pd2, MW = 2990



MBBF4 — EIGER and a TEM make a Synchrotron



MBBF4 — Data Accuracy (J. Holstein, TU Dortmund)

Structure of MBBF4• R1 = 22.7%(2941Fo > 4σF)

• R1 = 27.2%(4832Fo)

• GooF = 1.5• 564 parameters, 1054 restraints

After addition of hydrogen atoms and re-straints: Dual conformation of BF4 becomesvisible.

Structure refinement by J. Holstein



9 - Electron Crystallography of Macromolecules



Protein Crystals in the TEM

Lysozyme, ≈ 2.1Å resolution(Clabbers et el. (2017))

Thermolysin:

≈ 2×1×very thin µm3

Solvent reduces contrast(sample courtesy I. Schlichting)

Thermolysin:

about 3Å resolution



The Question of Resolution

e− X–ray resol. ratiodmin PDB-ID dmin PDB-ID d(e−)/d(X-rays)

Lysozyme 1.80 5K7O 0.94 1IEE 1.9Catalase 3.20 5GKN 1.50 1DGF 2.1Proteinase K 1.75 5I9S 0.83 2PWA 2.1Xylanase 2.30 5K7P 0.97 3AKQ 2.3Thaumatin 2.11 5K7Q 0.90 5X9L 2.3Trypsin 1.70 5K7R 0.75 4I8H 2.2Thermolysin 2.50 5K7T 1.12 5JVI 2.2

Liu et al., “Atomic resolution structure determination by the cryo-EM method MicroED”, Protein Science (2017), 26, 8–15

• Organic compounds reach similar resolution with electrons as with X–rays

• Proteins seem to be consistently 2× worse



High resolution data collection for electron Diffraction

• X-rays: Test crystals (Thaumatin, Lysozyme, . . . ) easily diffract to 1.2–1Å

• Electrons: about 2x worse so far

• Idea (I. Schlichting, K. Diederichs, independently): Combine serial crystallography with rotation method

1. Rotate sample at high dose with short lifetime but maximum resolution, e.g. 5◦ per crystal

2. Combine data from many crystals for data completness

• Very important open question in electron crystallography



10 - Preferred Crystal Oritentation & the Missing Wedge Problem



Missing Wedge in Electron Diffraction

Cu

Cu

140°

(a)

64°

Cu

Cu

(b)

• Crystals very often have a flat shape: always the same orientation

• Sample support stabilised by Cu-grid

• Copper grid too thick: intransparent for electrons

• Limited rotation range



Effect of Missing Data on Map and Structure

(e) (f)

10°

(g)

30°

(h)

50°



Complete Data from 3D Structured Grids - Coiled carbon grids

Brush Stroke causes carbon layer to coil



Complete Data from 3D Structured Grids - Nylon Fibres

Nylon fibres (≈ 100nm diameter) disturb preferred orientation

3D structured grids break

the preferred orientation of the crystals.

Complete data from < 5 crystals.



11 - Acknowledgements

• R. Pantelic (PSI/DECTRIS), S. De Carlo (DECTRIS), C. Zaubitzer (ScopeM), J. Wennmacher (PSI), J. Hol-stein (TU Dortmund), J. Heidler (PSI), A. Fecteau–LeFebvre (C–CINA), K. Goldie (C–CINA), E. Müller(PSI),S. Handschin (ScopeM), H. Stahlberg (C–CINA), N. Blanc (ScopeM), C. Schulze–Briese (DECTRIS)

• B. Luethi (DECTRIS), L. Wagner (DECTRIS), L. Piazza (DECTRIS), D. Mayani (DECTRIS)

• Y. K. Bahk (ETHZ), I. Regeni (TU Dortmund), T. Li (ETHZ), L.Muskalla (Uni Konstanz), A. Pinar (PSI), J.A.van Bokhoven (PSI/ETHZ), G. Clever (TU Dortmund)

• G. Santiso–Quinones (Crystallise!), G. Steinfeld (Crystallise!), R. Mezzenga (ETHZ), U. Shimanovich (Weiz-mann Inst.), I. Adrianssens Martiel (PSI), I. Schlichting (MPI Heidelberg), K. Diederichs (Uni Konstanz), J.Lübben (now Bruker AXS), M. Clabbers (Uni Stockholm)

nanoArgovia A3EDPI SNF Project 169258


Documents

Organic and macromolecular crystallography with electrons€¦ · Tim Grüne Electron Crystallography nanoArgovia A3EDPI Spot Intensity • Spots intensity depends on physics of interaction