Electron microscopy

Imaging is the

representation or reproduction of an objects form

Microscope magnifies an image of

an object that is too small to see easily

A telescope magnifies an image that

is too far away to see easily

Why high-energy electrons?

Their short wavelength allows small things to be resolved ed

Image formation comes from the

difference in density between biological material and the solvent

When water is cooled rapidly

it doesn’t form an ordered structure → preserves biological structure

Cooling water with liquid ethane

Needs large surface area to volume ratio by having thin samples → creates thin sample that electrons pass easily for imaging

Critical parameters for grid preparation

Humidity, blot force, and temperature

Images are noisy even with sensitive detectors to get good resolution you need

to cancel noise and build signal

To get a good resolution of 2D images of 3D objects

Need to reconstruct different 2D views into a single 3D map

Single particle analysis is based on

Aligning particles together

When working in 2D to create 2D images we have three variables

Translation in x & y, ‘In-plane rotation’

What can 2D classes tell you

First detailed look at your target, contaminants, symmetry, conformational changes, dense cofactors

In 3D process is the same as in 2D but there’s now 5 variables

2 translation x& y, 3 angles (psi, theta, phi)

Classification

Sorting particles into different classes based on their states

Biological molecules are

Heterogenous → sometimes biologically relevant, sometimes not

Classification allows you to sort

Particles into different classes to separate out this heterogeneity

Refinement (expectation step)

Compare each particle to a reference and translate and rotate

Refinement (maximisation step)

Average particle images together to improve the signal and cancel out the noise

Refinement → if resolution doesn’t improve

Continue

Refinement → is resolution stops improving

Stop

Classification (expectation step)

Compare each particle to a reference and translate and rotate

Classification (maximisation step)

Average particle images together to improve the signal and cancel out then noise

Classification (steps after maximisation)

Move particles to the class that matches the best → keep going around until user intervenes

Catalytic cycle of ATP synthase

Protons move through the F0 domain, rotating an axle connected to the F1 domain.

F1 domain is held by stator, the axle pushes on the F1 domain: drives the phosorylation od ADP with Pi to make ATP

3D classification of ATP reveals

rotary catalytic cycle of ATP synthase

Using a stable ATP synthase to resolve catalytic subsets

Allows deep classification to high resolution

Polytomella (an algae)

ATP synthase has a elaborated → much more stable

Filaments in biology

Often have a helical nature

A helix is formed by

Combining rotational symmetry with translational symmetry

Why are helical filaments helpful for cryo-EM analysis

Need to average over many particles in many different views, filaments provide all these views.

Pack asymmetric units into close proximity

Geometric relationship between asymmetric units is the same for whole filaments

Defining symmetry for helices

Define everything along ‘cylindrical coordinates’

Two key numbers that define a helix

Rise and twist

Rise

distance each subunit takes up along filament length

Twist

degree of rotation each subunit moves around the helix

Reconstructing filaments

Align each segment of the filament and find the values of twist and rise

Translations in x & y. x is set along the filament length and is incremented as rise

Tilt

defines whether looking along or side-on to the filament: a filaments run along within ice then side views dominate

psi (in-plane rotation)

will be dominated by values along filament length

Typical workflow: Single-particle analysis

2D classification → 3D classification → Refinement to high resolution

Typical workflow: Helical reconstruct

2D classification → 3D helical classification → Helical refinement to high resolution

Amyloid filaments are dominated

by beta-sheets → stabilised by backbone interactions with side-chains sticking out

Many proteins lose native tertiary fold to enter the amyloid state because

Its stable and hard for cell to deal with

Filaments are

robust, isolated from cells as the ‘sarkosyl-insoluble’ fraction

Filaments formed from

tau protein

Once it was possible to get filaments into EMs

These were subject to imaging and diffraction

2003 > 2017

Arrival of sensitive and fast direct electron detectors, better sample robots for working with cryogenic-grids, more stable microscope optics

Two forms of filament

Straight and paired

Both types of filaments adopt

Similar folds but held together differently

Paired filaments held together by

GGG motif tightly together

Straight filaments appear partially held together

by density connecting lysine

Early days of structural biology targets were purified

without recombinant techniques

Vast majority of targets to be

addressed with NMR or x-ray crystallography are produced using recombinant DNA technology

Tiny sample requirements of cryo-EM has opened up

Opportunities to work with real targets isolated from the real organism of interest

Advantages of ex vivo structural biology

Allows one to really explore what matters: biological molecules as they are in the cell.

Avoid problems with assembly pathways for complicated complexes

Difficulties with ex vivo structural biology

Functional techniques often require plenty of material. Easy to explore function if you have sufficient material for crystallisation.

As sample preps get smaller for cryo-EM harder to do functional analysis at same time, many structures may be inactive

Purifying such small quantities is hard

Problem at both purification stage and grid preparation stage

Tomography

Various process used to produce a cross-sectional image of something by detecting the passage of em waves, electrons etc.

How to do tomography within an electron microscope

Take an image → rotate grid → take an image → rotate grid etc.

will typically do one image per degree of rotation. Cryptically if doing cryo-tomography, does limitations apply

Focused ion-beam miling

Even with energy filter, limit to sample thickness. Can’t make cells smaller, use a focused ion beam to thin sample for tomographic collection

Correlative light-electron microscope

CLEM combines chemical/genetic specificity of fluorescent probes/GFP with the ultrastructural information of EM.

Often uses fixed cells

In the long run should be able to combine with subtomogram averaging

cells are full of

Complicated mess → need to find target of interest

Subtomogram averaging

Similar to other reconstruction techniques: average homogenous elements to cancel noise and improve resolution.

Start with 3D volumes from tomograms rather than 2D images

Iterative process like single particle analysis & helical reconstruction

Subtomogram averaging reveals

Structure of an antibiotic-bound ribosome

How do electrons in an electron microscope behave?

relativistic velocity

Sensitive to magnetic fields

Highly energetic

Relativistic velocity

Takes about 5 ns for electron to come down the microscope column. Pass sample in 0.5 fs

Sensitive to magnetic fields

Allows the use of electromagnetic lenses to focus and magnify

Highly energetic

Carrying huge amount od energy and can damage sample

How do electrons interact with the sample?

Electrons interact strongly with matter

Three main scattering interactions when meeting a sample

Three main scattering interactions

Non-scattered, elastically scattered, inelastically scattered

inelastically scattered

loses energy to sample change in vector

non-scattered

no change in energy, no change in vector

Elastically scattered

no change in energy change in vector

ratio of inelastic events to elastic events is

different for electrons and x-rays > implications for damage

Inelastically scattered electrons are bad in two ways

Loses energy → shorter wavelength → not focused by microscope → no good for imaging

Lost energy damages the sample

Direct electron detectors

detect electrons without any intermediate stage

For each productive elastically scattered electrons there are

3 inelastically scattered electrons: have to make them count > critical to have a sensitive detector

Motion

Tension builds in grids when they are frozen > relieved by electron beam > motion that degrades images > correcting this motion improves contrast

Damage weighting

During exposure, damage grows > affects different scales to different extents > get high res info from early frames but low res info from all

Transmission electron microscopes primarily produce

Phase contrast rather than amplitude contrast

Amplitude contrast comes from

electrons not exiting the optical system: scattered off optical axis or absorbed by energy filter

Phase contrast comes from

the interaction of non scattered and elastically scattered beam

Amount of phase contrast depends on

Shift of elastically scattered wave relative or non scattered > change by defocusing lens > how image affected depends on defocus

Condensor system

Three vs four lens system

Increases flexibility for illuminated area whilst maintaining parallel illumination

Energy filter

Either sits in column or after column

Critical for thicker samples, nice for single-particle/filament work

Detector

For cryogenic work, all detectors are now direct electron detectors and use electron-counting

Newer generations are faster & more sensitive

Electron source (microscope hardware)

Filament vs field-emission gun

Sample handling system

Side entry (slow, cheap, manual) vs autoloader (fast, expensive, automated)