What actually happens:
According to current understanding of the freezing of
water, whenever it is done quickly there will be some degree of supercooling of
the water and the first solid phase formed has a random arrangement of layers in
one dimension as referenced in my post of January 2025 (see Malkin et al. 2015).
A second, slower step then occurs to convert the partially random solid into
regular hexagonal ice. Because the rate of formation of the first, partially
random solid is limited by the rate of diffusion of latent heat away form the
solid/liquid interface, it grows in a dendritic (branching) and fractal
(repeating at ever smaller scales) form. Think of a snowflake growing in 3
dimensions, branching at ever smaller scales until all liquid space is solid.
This is the usual mode of solid growth when the process is limited only by
diffusion. The first form of ice is transparent to an electron beam whereas
hexagonal ice is opaque under the beam and shows characteristic patterns. Where
cooling isn’t quick enough, the cryoelectron microscopist sees only hexagonal
ice. Where cooling has been quicker the microscopist sees the transparent
metastable form of ice. There is no remaining liquid and no remaining interface.
What Dubochet et al. hoped would happen:
The latent heat of freezing becomes less as the temperature of freezing decreases.
There is a thermodynamic explanation that we don’t need here. It is enough to know
that from measurements from zero Celsius downwards a few degrees we can
extrapolate to a temperature where there would be no latent heat of freezing.
This would be in the vicinity of -80 Celsius. In this hypothetical situation
the rate of transformation from liquid to solid would not be limited by flow of
heat away from the growing interface. It would be possible for the whole sample
to simultaneously transform from liquid to solid. If it were possible to supercool
water to about -80 Celsius, the resulting solid might retain the structure of the
liquid from which it formed and any objects that were in the liquid might be held
in the same form they had in the liquid. The idea was that there would be a liquid
to solid transformation without the formation of a (troublesome) interface or with
such a rapid spread of the interface that all structure would be preserved.
What happens in other branches of science:
In many areas of science and technology there had been decades of research
exploring the limit to the supercooling of water. What Dubochet et al. wanted to
achieve had, of course, been much sought after for the freezing of food. Food
scientists found it impossible to cool water below maybe -30 to -40 without it
freezing. At this temperature freezing causes heat release which slows the freezing.
There is no such thing as instant freezing in food technology. In aviation there
was and is a recognised hazard of supercooled water droplets in clouds solidifying
on wings causing loss of lift and extra weight. Experiments and experience showed
that this only occurs at low to medium altitudes corresponding to temperatures
down to -20 to -40 Celsius. At lower temperatures all cloud particles are ice and
not inclined to adhere to wings. More recently atmospheric physicists modelling
the heat flows through the atmosphere have seen the same result. There are very
good explanations of icing in aviation online. Here is a good example
www.weather.gov/source/zhu/ZHU_training_Page/icing_stuff/icing/icing.htm
Comments
We don’t know whether Dubochet et al. were aware of this knowledge when they
started the work on cryoEM but it seems it was brought up by a reviewer before
publication. According to Dubochet’s own account of the initial publication,
there was a reviewer criticism that the result shouldn’t be possible. It took
considerable effort by the authors (in my opinion maybe helped by their
supporter Sir John Kendrew) to get the work published. People were persuaded
that the apparent random orientation of added particles (viruses or proteins)
showed that liquid water had been instantly “vitrified” into a glasslike form of
ice. In fact random orientation is also the expected outcome if suspended
particles adsorb to the growing dendritic interfaces described at the start of
this post.
What it means for protein quaternary structure:
From the voluminous literature in food technology and atmospheric physics we
can say that the temperature at the interface during freezing is close to zero
Celsius. We can also estimate the time it would take for a protein to be engulfed
by ice to be in the order of a millisecond. From molecular dynamics studies we
can estimate the time it would take for a protein to reorganise its quaternary
structure to be of the order of a microsecond. Plenty of time for change.
Atmospheric physics has also measured the ice/water surface energy and found it
comparable to the general solid/water surface energy. We can say that if a protein
is one whose quaternary structure may be affected by adsorption to an interface,
it is likely to take up the adsorbed quaternary structure during cryoEM just as
it is during crystallisation and just as it is before transmission EM. All these
techniques share a common uncertainty.
