π Health and biotech π Science and technology
Biomolecules: three techniques at the cutting edge of research

Cryogenic electron microscopy: the resolution revolution?

Pierre-Damien Coureux , Assistant Professor in Electron Microscopy at BIOC* at École Polytechnique (IP Paris)
On April 20th, 2022 |
5 mins reading time
Cryogenic electron microscopy: the resolution revolution?
Pierre-Damien COUREUX
Pierre-Damien Coureux
Assistant Professor in Electron Microscopy at BIOC* at École Polytechnique (IP Paris)
Key takeaways
  • Modern structural biology was born in the 1950s with the first publication of the DNA double helix structure and the first structures of proteins, which were awarded two Nobel prizes in 1962.
  • Although the first electron microscopes existed well before the 1950s and made it possible to observe materials at high resolution, for a long time the molecules of living organisms remained difficult to observe with this technique. In biology, this structural approach was thus often considered to be a big magnifying glass.
  • Over the last ten years or so, the technique has made revolutionary advances, allowing molecules to be seen at the atomic scale and as small as haemoglobin.
  • If Covid had appeared 15 years ago, researchers would never have been able to obtain its structure so quickly.
  • The next challenge is to see the interior of cells in sufficient detail to locate all the known atomic models of life and to understand their interactions.

“To under­stand, we must observe,” that’s the mot­to of mod­ern struc­tur­al biol­o­gy. It is a field that aims to obtain the struc­ture of bio­log­i­cal objects to bet­ter under­stand the role of a par­tic­u­lar mol­e­cule in a cell. This mod­ern struc­tur­al biol­o­gy was born in 1953 with the pub­li­ca­tion of the struc­ture of the DNA dou­ble helix by Wat­son and Crick1 and a few years lat­er with the first pro­tein struc­tures by Kendrew and Perutz. Since then, struc­tur­al biol­o­gy has pro­vid­ed researchers with valu­able infor­ma­tion for under­stand­ing and pro­tect­ing life.

A complex history

Tech­niques avail­able for resolv­ing the 3D struc­ture of bio­log­i­cal objects become avail­able and evolved in turn. X‑ray crys­tal­log­ra­phy, already mas­tered in the 1910s, under­went a rev­o­lu­tion in the 1990s thanks, among oth­er things, to the use of crys­tal cryo­gen­ics and the qual­i­ty of syn­chro­trons that gen­er­ate X‑rays. Nuclear mag­net­ic res­o­nance, dis­cov­ered by chance in the 1940s, emerged in struc­tur­al biol­o­gy in the 2000s and is still used today, espe­cial­ly to study the dynam­ics of pro­teins. For a long time, elec­tron microscopy (which was born in the 1930s) was con­sid­ered in biol­o­gy as a pair of binoc­u­lars to see the infi­nite­ly small but with­out pro­vid­ing high-res­o­lu­tion struc­tur­al information.

The prob­lem: the mol­e­cules of liv­ing organ­isms are minis­cule. For exam­ple, the SARS-CoV­‑2 Spike pro­tein weighs ~180–200 kDa2 – equiv­a­lent to three haemo­glo­bin mol­e­cules3. Elec­tron microscopy images have a lot of noise due to the inter­ac­tion of the elec­tron beam of the micro­scope with the sam­ple. As a result, noth­ing can be seen on the images if the object observed is too small. It is only in the last ten years or so that a rev­o­lu­tion has tak­en place. The design of micro­scopes has been improved. Also, the qual­i­ty and speed of the detec­tors used have made it pos­si­ble to record bet­ter qual­i­ty images, the soft­ware used to help process the data is more intu­itive and more effi­cient, and the speed of cer­tain cal­cu­la­tions has been improved by a fac­tor of 50 thanks to com­put­er graph­ics cards. If the SARS-CoV­‑2 pan­dem­ic had appeared some 15 years ago, we would nev­er have been able to obtain its struc­ture so quick­ly and so precisely!

Better and faster

Elec­tron microscopy orig­i­nat­ed in Ger­many and was first used in physics. Biol­o­gy ben­e­fit­ed from devel­op­ments in physics and the first images of bio­log­i­cal sam­ples date from the 1950s. It was not until 1968 that the Amer­i­cans Rosier and Klug demon­strat­ed that 2D images tak­en with an elec­tron micro­scope could be used to trace the 3D struc­ture of the object stud­ied. The first struc­ture of a mem­brane pro­tein came from a bac­teri­um. It dates back to 1975 and was made by British pio­neers Unwin and Hen­der­son (see pho­to below). Hen­der­son received a Nobel Prize in 2017 for his pio­neer­ing work over 40 years with researchers Frank and Dubochet.

First mem­brane pro­tein struc­ture obtained in 1975 at 7Å res­o­lu­tion, a record for the time4.

It took anoth­er 40 years to obtain the almost-atom­ic res­o­lu­tion (3Å aver­age res­o­lu­tion) of the human ribo­some (made up of sev­er­al strands of RNA and about fifty pro­teins). Today, more than 10 struc­tures per day are deposit­ed in researcher data­bas­es and more than 20% of them are high-res­o­lu­tion data.

High res­o­lu­tion human ribo­some mod­el first obtained in 20155.

As ear­ly as Jan­u­ary 2022, the pre­cise struc­ture of the Spike pro­tein of the omi­cron vari­ant of SARS-CoV­‑2 (which only appeared in Novem­ber 2021) and its human tar­get (the ACE2 lung recep­tor) was revealed. This has led to new strate­gies to fight more effec­tive­ly against the infec­tion of this virus in our lungs6. Also, knowl­edge of the exact arrange­ment of the amy­loid fibres involved in Alzheimer’s dis­ease has made it pos­si­ble to design drugs that slow down the accu­mu­la­tion of these fibres in our brain7. Thus, more and more bio­log­i­cal mod­els are being revis­it­ed and rapid­ly improved thanks to high-res­o­lu­tion 3D struc­tures obtained with the lat­est advances in elec­tron microscopy. Tech­no­log­i­cal bar­ri­ers are open­ing up to now observe pro­teins alone or in com­plex with oth­er part­ners to bet­ter under­stand their role in the cell. No won­der the phar­ma­ceu­ti­cal indus­try is invest­ing heav­i­ly in this tech­nique to screen their new ther­a­peu­tic mol­e­cules more eas­i­ly and quickly!

The limits of the technique

An elec­tron micro­scope oper­ates in an ultra-high vac­u­um (10-8 mbar) so that the elec­trons flow­ing through it gen­er­ate as lit­tle extra­ne­ous noise as pos­si­ble. The only way to observe a bio­log­i­cal sam­ple is to trans­form it into a sol­id ice cube (by vit­ri­fi­ca­tion) so that the sam­ple, once in the micro­scope, remains in sol­id form while retain­ing its orig­i­nal shape: this is what is known as cryo-EM!

Image of an elec­tron microscopy grid tak­en at low mag­ni­fi­ca­tion. You can see the cop­per frame on which the thin per­fo­rat­ed car­bon mem­brane rests: this is the sup­port for any bio­log­i­cal sam­ple stud­ied by cryo-elec­tron microscopy8.

The sup­port on which the sam­ple is deposit­ed is, in gen­er­al, a per­fo­rat­ed car­bon mem­brane of about 10 nm thick­ness which rests on a cop­per frame­work form­ing tiles of about 100x100 µm. The sam­ple will be trapped in the holes of the per­fo­rat­ed mem­brane and the most crit­i­cal step is to form a thin film in each hole. The cop­per frame­work is a stan­dard 3mm diam­e­ter grid. This grid is then placed in the elec­tron microscope.

The elec­tron beam used (gen­er­at­ed by the elec­tron micro­scope) will pass through the sam­ple on the grid. The images record­ed by trans­mis­sion will pro­vide infor­ma­tion on the 3D organ­i­sa­tion of the atoms that the beam encoun­ters, but they also gen­er­ate a lot of noise. This is why it is nec­es­sary to record thou­sands of images to aver­age the infor­ma­tion and obtain a good sig­nal-to-noise ratio, in order to unam­bigu­ous­ly deter­mine the shape of the bio­log­i­cal objects studied.

In each image, a few dozen bio­log­i­cal objects (called par­ti­cles) can be seen. The shapes observed in the images cor­re­spond to dif­fer­ent pro­jec­tions of the same 3D object. A pro­jec­tion con­tains all the struc­tur­al infor­ma­tion to trace the atom­ic coor­di­nates of the 3D object. The images must then be analysed to match the rota­tion angles applied to the orig­i­nal 3D object, to obtain the observed pro­jec­tion in the micro­scope. By col­lect­ing a large amount of data, it is pos­si­ble to obtain the 3D struc­ture of a bio­log­i­cal object on an atom­ic scale. So, research that used to take sev­er­al months now takes only a few weeks or even days.

Not every bio­log­i­cal sam­ple can be observed with an elec­tron micro­scope yet. At present, objects small­er than 20–25 kDa (about 3 nm) are very dif­fi­cult, if not impos­si­ble, to observe. But this is where the oth­er struc­tur­al approach­es described at the begin­ning can fill this gap!

One of the lab­o­ra­to­ries at École Poly­tech­nique (BIOC) is a pio­neer in its field, using the elec­tron microscopy approach to study, among its var­i­ous research top­ics, the ribo­somes of Archaea. The pre­lim­i­nary data for the work pub­lished910 were car­ried out at CIMEX (see box).

Another revolution on the way? 

Over the past decade, major rev­o­lu­tions have tak­en place in the design of micro­scopes, in the cam­era tech­nol­o­gy used to take images and in the soft­ware used to process these images. This dis­ci­pline has been brought up to date and has rad­i­cal­ly changed the way researchers approach bio­log­i­cal problems.

The next step is to see the mol­e­cules in their cel­lu­lar con­text. Elec­tron tomog­ra­phy is still a tra­di­tion­al method that requires a lot of know-how and is not very pop­u­lar. The prepa­ra­tion of the sam­ples is even more crit­i­cal: it requires us to be able to cut a thin lamel­la of about 200 nm thick­ness out of a cell ice cube. This slide is then observed at dif­fer­ent angles in the micro­scope to cal­cu­late a 3D map of the region of inter­est at medi­um res­o­lu­tion. Com­put­ers are then able to recog­nise pro­files of mol­e­cules in this 3D map to recon­struct a mod­el of the 3D map. There are many devel­op­ments in this field and in a few decades it should be pos­si­ble to look at a par­tic­u­lar pro­tein any­where in the cell in its cel­lu­lar con­text. The nano-uni­verse of cells will soon hold no secrets for us…


The Cen­tre Inter­dis­ci­plinaire de Micro­scopie Elec­tron­ique de l’É­cole Poly­tech­nique (CIMEX) is a plat­form hous­ing sev­er­al micro­scopes where physi­cists, chemists and biol­o­gists can meet and use high-per­for­mance elec­tron micro­scopes. One of the micro­scopes, called NanoMAX, is a world pro­to­type. It is used by physi­cists and chemists to study, among oth­er things, the growth of car­bon nan­otubes at high res­o­lu­tion and in real time. The oth­er micro­scope used more by biol­o­gists, called Nan’eau, is ver­sa­tile and well-equipped. It allows impor­tant pre­lim­i­nary data to be obtained before mov­ing on to state-of-the-art micro­scopes (avail­able at 3 nation­al ref­er­ence cen­tres). The images then record­ed will be of suf­fi­cient qual­i­ty to obtain the high-res­o­lu­tion 3D struc­ture of the object studied.

4Hen­der­son, R., Unwin, P. Three-dimen­sion­al mod­el of pur­ple mem­brane obtained by elec­tron microscopy. Nature 257, 28–32 (1975)


Pierre-Damien COUREUX

Pierre-Damien Coureux

Assistant Professor in Electron Microscopy at BIOC* at École Polytechnique (IP Paris)

After a PhD in structural biology on myosin molecular motors at Institut Curie in 2004, he left for the United States to study bacterial and plant photoreceptors using several approaches including electron microscopy. He was recruited in 2008 at the Ecole Polytechnique and works mainly on the cellular machines that synthesise our proteins: the ribosomes, with those of the Archaea as a study model. He was appointed head of CIMEX in 2020 and is also involved in teaching biology at X.

*BIOC: a joint research unit CNRS, École Polytechnique - Institut Polytechnique de Paris