J. Cell Sci. is, 693-701 (1974)
693
Printed in Great Britain
DECREASE IN SECTION THICKNESS ON
EXPOSURE TO THE ELECTRON BEAM; THE
USE OF TILTED SECTIONS IN ESTIMATING
THE AMOUNT OF SHRINKAGE
PAULINE M. BENNETT
MRC Muscle Biophysics Unit,
26-29 Drury Lane, London WCzB 5RL, England
SUMMARY
Tilting experiments on sections of embedded light meromyosin paracrystals of known periodic
structure indicate that the section thickness decreases during examination in the electron
microscope. The material remaining in the section after irradiation collapses uniformly to about
one half the initial thickness. This must be taken into account when measurements are made
on tilted sections.
INTRODUCTION
It has been known for some time that considerable changes in the physical properties of thin sections may be induced by irradiation in the electron microscope. There
have been few systematic studies, however, of the nature of these changes, particularly
when the sections contain biological material. Cosslett (1961) showed that thin sections
of several embedding media commonly used in the electron microscopy of biological
systems became thinner by a finite amount after exposure to the electron beam at
normal viewing intensities. The change in thickness was between 20 and 50 %,
depending on the embedding medium, and was found to be very rapid for the beam
intensities used. It is also known that when supportingfilmsand sections of embedding
media are exposed to an electron beam much of the material evaporates and a residue
containing a high proportion of carbon remains; in some cases more than half the
original mass may be lost (Bahr, Johnson & Zeitler, 1965; Reimer, 1965). The loss of
weight can be correlated with the decrease in section thickness.
Recently Willis (1973) has shown that the thickness of a section of biological
material changes dramatically on exposure to the electron beam at normal viewing
intensities for high-resolution microscopy. By re-embedding a section, part of which
had been irradiated, and then sectioning perpendicular to the original section, he
showed that the irradiated area was only half as thick as the non-irradiated region.
Such a change in section thickness will cause distortions in the 3-dimensional
structure of the specimen under observation and although this will have little effect on
measurements made on sections viewed normally, it is clear that measurements made
on tilted sections will require correction. The use of goniometer stages for tilting
sections has become a widely used technique for investigating 3-dimensional structures.
44-3
694
P- M- Bennett
However, to obtain reliable measurements it is necessary to know the amount of
shrinkage and whether the shrinkage is uniform throughout the section.
The work to be reported here developed from a study with my colleague Dr G.
Offer on the 3-dimensional structure of light meromyosin (LMM) paracrystals. LMM
is a proteolytic fragment derived from the tail of the muscle protein myosin. The
molecule is a coiled coil a-helical rod approximately 70 nm long. At low ionic strength
these molecules aggregate to form paracrystals, the largest of which can be seen in the
light microscope (Szent-Gyorgyi, 1953). In the electron microscope the paracrystals
show a well defined periodic structure either when negatively stained (Philpott &
Szent-Gyorgyi, 1954; Szent-Gyorgyi, Cohen & Philpott, i960) or embedded, sectioned
and stained (P. M. Bennett & G. W. Offer, in preparation). I report here how measurements of the main axial periodicity of the paracrystals made on micrographs of tilted
sections differed significantly from those made on micrographs of untilted sections, the
difference increasing with the angle of tilt. The results indicate that the sections shrink
to about half their initial thickness on exposure to the electron beam, and this shrinkage appears to be uniform throughout the thickness of the section.
The purpose of this paper is to show that measurements made on tilted sections may
be smaller than the original spacing in the specimen before irradiation and how this
must be taken into account when deriving 3-dimensional structures from a tilted
series of micrographs.
MATERIALS AND METHODS
L M M was prepared by the method of Szent-Gy5rgyi etal. (i960) and dialysed against 0-15 M
KC1, 10 mM potassium phosphate buffer, pH 6-5 to form paracrystals. The paracrystal suspension
was spun at 1000 g for 15 min to form a loose pellet which was fixed in 3 % glutaraldehyde in the
same medium for 3 h at 4 °C, postfixed in 2 % osmium tetroxide in 1 1 2 % sucrose, o-15 M KC1,
10 mM potassium phosphate buffer, dehydrated in an ethanol series and embedded in Araldite.
Sections of thicknesses varying from 40 to 150 nm as estimated by their interference colours
(grey to purple-blue) were cut on an LKB microtome and mounted on grids which had been
coated with collodion followed by carbon. The sections were stained with 4 % uranyl acetate
followed by Reynolds lead citrate.
For comparison with the positively stained sectioned material a whole paracrystal preparation
was made by negatively staining a drop of the original paracrystal suspension on a grid with 2 %
potassium phosphotungstate at pH 4 0 . The excess stain was washed off with o-i M ammonium
acetate, leaving a positively (i.e. specifically) stained structure.
Observations on tilted sections were carried out in a Philips EM 300 microscope fitted with a
goniometer stage. The microscope was operated under normal working conditions at 80 kV,
with an anti-contamination device cooled by liquid nitrogen. No special precautions were taken
to use low beam intensities since it was necessary to observe the change in appearance of the
specimen at high magnification when tilting. An area of interest in the specimen was thereby
exposed for up to a few minutes to a beam current of approximately io~* A cm" 1 . In order that
the final measurements would not be affected by compression during sectioning the following
procedure was used. (1) The section was located in the microscope and the grid rotated until
the tilt axis lay parallel to a knife mark in the section. (2) In any section there is a random array
of paracrystals some of which have their long axis perpendicular to the tilt axis. These could be
tilted until their inherent periodicity was seen clearly. (3) All the required measurements were
made perpendicular to the tilt axis and hence to the knife mark. Micrographs were taken usually
at a magnification of x 30000. The microscope was calibrated using a diffraction grating replica
with either 28800 or 54864 lines in." 1 . The periodic repeat in the paracrystals was measured on
the original micrograph with a travelling microscope.
Irradiation changes and tilted sections
695
RESULTS
LMM forms several types of polymorphic aggregates (King & Young, 1970;
Chowrashi & Pepe, 1971; Katsura & Noda, 1973). Under the conditions used here the
predominant form is that of long spindle-shaped paracrystals, all of which when
stained exhibit an axial periodicity of approximately 40 run. Fig. 4A shows part of a
whole positively stained paracrystal; 40-nm banding can be seen throughout but the
repeating pattern is clearer near the thin edges of the paracrystal.
Sectioned paracrystals are shown in Fig. 4B-F. In any one section all orientations of
the paracrystals are seen. Some paracrystals have been fortuitously sectioned parallel
to their long axis and show a sharp periodic band pattern (Fig. 43) similar to that seen
in the whole paracrystal of Fig. 4A. Other paracrystals seen in the same section also
show a repeating pattern but the bands are wide, with diffuse edges (Fig. 4c). On
tilting the section about an axis perpendicular to the long axis of such a paracrystal the
appearance of the diffuse bands changes, becoming more diffuse on tilting in one direction and sharper on tilting in the opposite direction. When the bands become sharp the
appearance of the paracrystal is similar to that of the paracrystals whose bands are seen
sharply without tilting (compare Fig. 4D with 4B). The error in estimating the tilt
angle at which the bands were sharpest depended on the section thickness. The results
were obtained from sections of 2 thicknesses, 80-90 nm and 120-150 nm, as estimated
by their interference colours after cutting. In the thicker sections the uncertainty
in the tilt angle for sharpest banding at small tilt angles was approximately + 6°,
whereas at large tilt angles it was approximately ±3°. Although using the thicker
sections is more accurate at low angles, at high tilt angles the image is not so clear and
the thinner sections are more useful.
This evidence implies that the sharply staining lines represent planes of stain
passing right through the paracrystal, which when viewed edge on appear as sharp
narrow lines, but when viewed at an angle appear as wide bands with diffuse edges
(Fig. 4c and 4D). The distance between these planes of stain in both the positively
stained whole paracrystal preparation and the sectioned paracrystals showing sharp
bands without tilting is approximately 40 nm. The same value would be expected in
sharply banded tilted paracrystals, since it is the perpendicular distance between planes
that is measured. This is not the case. Fig. 4D-F shows sections of paracrystals at the
same magnification which have been tilted through different angles in order to show a
sharp banding pattern. It can be seen that the repeat distance is smaller for those paracrystals requiring large tilts. For example, if the angle required was 6o° the measured
spacing was only 25 nm. This decrease in the periodicity with tilt angle was observed
whether positive or negative tilts were required. The measurements obtained from
several paracrystals from 4 sections, 2 at each thickness, are collated in Fig. 1. The
experimental points show the repeat distance q (measured after tilting) plotted against
the angle of tilt. The 4 symbols refer to the 4 sections used, and the solid lines are
theoretical curves which will be discussed later. Clearly the measured periodicity
varies systematically with the angle of tilt required, and this variation is independent
of section thickness, since the results for the 2 thicknesses used do not differ signifi-
696
P. M. Bennett
cantly. (Compare the triangles with the circles in Fig. 1.) This latter observation
suggests that fractional shrinkage in sections of different thicknesses is the same and
that surface effects do not contribute.
0
10
20
30
40
50
60
Fig. 1. Plot of the periodicity versus tilt angle. Experimental points show the measured
periodicity, in ran, q in paracrystals when sharp banding was seen plotted against the
angle of tilt, in degrees, <j> necessary to reveal such sharp banding. T h e 4 symbols correspond to measurements made on different sections. A , A , original section thickness
150 / t m ; O, • , original section thickness 80 /im. T h e solid lines are theoretical
curves calculated from equation (3) in text with different values of section s h r i n k a g e / .
/ = 1 corresponds to no shrinkage; / = 0 7 , to 3 0 % shrinkage; / = 0 5 , to 5 0 %
shrinkage; a n d / = 0-3, to 7 0 % shrinkage.
One possible explanation of these results is that there was a magnification change
during tilting. This is unlikely since, in the goniometer stage that was used, the height
of the specimen can be adjusted until the tilt axis passes through the section at the
region of interest and little focus change is necessary over a wide range of tilts. However, in order to check this possibility the periodicity in a given paracrystal was measured at a number of tilt positions either side of zero tilt whether the stripes were
fuzzy or sharp. The periodicity was found to vary as expected with the cosine of the
angle of tilt. Therefore there was no significant change in the magnification on tilting.
The results shown in Fig. 1 can be explained if it is assumed that the section gets
thinner by contracting on to the grid. Evidence that the section is indeed less thick
after irradiation than originally estimated comes from pairs of micrographs of the same
paracrystal, one at zero tilt with diffuse bands and the other showing sharp bands after
Irradiation changes and tilted sections
697
tilting. This can be seen by reference to Fig. 2, which shows a section of original thickness a through a paracrystal. The plane of section is originally at some arbitrary angle
(6) to the long axis of the paracrystal. The solid oblique lines represent the planes of
stain passing through the paracrystal. When such a section is viewed normally the
planes of stain will show as wide stripes of width a tan 6, but after tilting the section
through an angle 0 they will appear as sharp lines. An estimate of section thickness can
therefore be obtained, given the width of the diffuse stripes at zero tilt and the angle
of tilt necessary to see sharp banding. In Fig. 4c the width of the dark stripe in the
paracrystal is approximately 20 nm and the angle of tilt necessary for sharp banding is
Electron beam
Reduced
Original
section
thickness
Fig. 2. Diagram to show the effect of electron irradiation and resultant change in
section thickness on the 3-dimensional structure of LMM. A section of original
thickness a passes through a paracrystal at an angle 0 to the long axis of the paracrystal. The solid oblique lines represent the planes of stain which periodically traverse
the paracrystal perpendicular to the long axis. Viewed normally this sectioned paracrystal would show diffuse banding but on tilting through 0° a sharp banding pattern
would be seen of periodicity p. Subsequent to irradiation the section shrinks to a
thickness / . a, and the original planes of stain now become the dotted oblique lines,
and the paracrystal exhibits a new periodicity q which can be seen after tilting
through <j>°.
33° (Fig- 40), indicating a section thickness of approximately (20/tan 330) x 35 nm.
However, this section when cut was judged by its interference colour to be 80 nm
thick. Therefore the section during irradiation has suffered a change of thickness
greater than 50 %. Similar shrinkage was deduced from other pairs of micrographs.
These estimates, however, are not very accurate because of the errors involved in
estimating the width of the diffuse bands and the original section thickness.
The simplest assumption (see Discussion) which will explain the results in Fig. 1
and the change in section thickness is that during exposure to the electron beam the
section shrinks in a direction transverse to the grid to a new thickness f.a shown in
Fig. 2 by a dashed horizontal line, where/is the fraction of the original thickness. The
original planes of stain now become the dotted oblique lines, and the section now has
to be tilted through a greater angle <f) in order to see sharp bands. In addition the
measured periodicity is now q, which is smaller than the original periodicity^) which
698
P. M. Bennett
was independent of 6. The expression for the new inter-plane distance q can easily be
derived in terms of the original spacing/), the fractional section thickness/ and the
angle of tilt <j>. From Fig. 2:
*
then, since cos2# + sin2# = 1,
(1)
costf
tan 6 = / t a n <j>;
(2)
q = p (cos2 0+/ 2 sin 2 0)i.
(3)
Note that q is independent of the original section thickness a, as was observed in the
experimental data, if/is constant. The solid lines in Fig. 1 gives values of q calculated
f=1
1500
-
1000 -
075
Fig. 3. The results and theoretical curves of Fig. 1 plotted as q1 versus sin'$
to give a linear plot.
from the above equation with/) = 41 nm and with values of/equal to 1 (no shrinkage),
°"7 (3° % shrinkage), 0-5 (50 % shrinkage) and 0-3 (70 % shrinkage). The experimental
data can be seen to follow the general shape of the theoretical expression. Most of the
experimental points fall within the limits of 30-70 % shrinkage and tend to cluster
around the line corresponding to 50 % shrinkage, which agrees with the rough estimate
derived earlier from the section thickness before and after irradiation.
Irradiation changes and tilted sections
699
For most purposes it is more convenient to have the data in a linear form. Equation
(3) can be rearranged to give
qZ=p*-p*(i-P)sm*<f>.
(4)
2
2
2
2
A plot of q versus sin (j> will give a straight line with intercept p on the q axis and
intercept (1 —Z2)"1 on the sin20 axis. Thus both p and/can be determined. Fig. 3 shows
the experimental data and the theoretical lines from Fig. 1 replotted with these
coordinates. This graph will be considered further in the discussion.
DISCUSSION
The only simple explanation of the experimental results presented here is a uniform
collapse of the material remaining in the section on irradiation. Estimates of section
thickness from measurements of the width of the diffuse bands in the paracrystals
and the angle of tilt necessary to make these diffuse bands sharp indicate that the
section is only half as thick after irradiation as originally estimated, in agreement
with the results of Willis (1973) and Cosslett (1961) at high beam-current densities.
This is normally the beam condition when a tilted series of high-resolution micrographs is required.
This change of section thickness does not appear to be a surface effect as might arise
from preferential evaporation of material from the surface or etching by air leaking in
at the specimen region, since examination of the micrographs indicated that the shrinkage is uniform throughout the thickness of the section. The lines of stain are sharp even
when seen at high angles of tilt. Distortion created by local differences in shrinkage
between the surfaces and the middle of the section would make the bands diffuse and
in particular the sub-banding would not be seen clearly. In addition, surface effects
would contribute with different weight to the results from sections of different thicknesses; but both thicknesses used showed the same relative decrease.
How does such a uniform decrease in section thickness occur? There is little or no
shrinkage parallel to the grid since the measured periodicity in sectioned paracrystals
whose bands are seen sharply without tilting is not significantly different from the
repeat seen in negatively stained whole paracrystals (i.e. approximately 40 nm). This
can be compared with a value of 43 nm derived from X-ray diffraction of LMM
fibres (Szent-Gyorgyi et al. i960). Presumably the section is prevented from shrinking
in this direction by its attachment to grid bars or the supporting film. The simplest
and most reasonable explanation for the change in thickness is that as the section loses
mass by evaporation it collapses in a direction transverse to the grid, since there is no
retaining force in that direction.
In this investigation only one kind of specimen has been used, LMM embedded in
Araldite, and it is not clear whether the paracrystalline regions shrink more or less
than the pure Araldite. Indeed the shrinkage factor may be different for other biological materials in Araldite or for LMM in different embedding media, since sections
of pure embedding media shrink by different amounts (Coslett, 1961). However, this
system of a specimen with a well defined periodic structure could be useful in investigating some of these shrinkage effects. Collagen fibrils, being more readily available
700
P. M. Bennett
than LMM, are an obvious example. An analysis as presented here of the periodic
repeat seen at different tilt angles would reveal the shrinkage factor and thereby
obviate the necessity of re-embedding and sectioning irradiated sections.
Changes in thickness cause distortion in 3-dimensional objects, and therefore one
should be aware of the dangers in making measurements on tilted sections. Shrinkage
in thickness causes 2 effects. One is the change in observed dimensions at high tilt
angles compared to those at low tilt. The original value can be calculated by geometrical considerations if the shrinkage is known or, if values are available at a variety
of different tilt angles, the real value can then be obtained by extrapolation to zero tilt
using the linear plot shown in Fig. 3. The other effect is the increase in tilt angle
generally necessary to see certain views of a structure. The angle of tilt required for a
given shrinkage factor can be calculated from equation (2). For example, if 2 orthogonal views of a structure are required (e.g. + 450 originally), with 50 % shrinkage it
would be necessary to take views at ±65°, a tilt range which is not normally possible
on standard goniometer stages.
The loss of mass of an irradiated section is very rapid and terminal section thickness
is achieved 'instantaneously' (Coslett, 1961). More specifically, Willis (1973) showed
that shrinkage has already occurred by the time one high-resolution micrograph has
been taken. Since tilting experiments take longer than this it is clear that shrinkage
must always be taken into account when making quantitative measurements on tilted
sections.
I thank Mrs Diana Terry for technical assistance, Mr Z. Gabor for photographic assistance,
Dr J. Koretz for computing the theoretical curves, and Dr A. Elliott, the late Professor Jean
Hanson, and Drs E. O'Brien, G. Offer and R. Willis for discussion.
REFERENCES
G. F., JOHNSON, F. B. & ZEITLER, E. (1965). The elementary composition of organic
objects after electron irradiation. Lab. Invest., Suppl. 14, 377-395.
CHOWRASHI, P. K. & PEPE, F. A. (1971). Some aspects of light meromyosin aggregation. In
First European Biophysics Congress, Baden, Austria, pp. 429-436. Verlag der Wiener Medizinischen Akademie.
COSSLETT, A. (1961). The effect of the electron beam on thin sections. In Proc. Europ. reg.
Conf. Electron Microsc, Delft i960, vol. 2, pp. 678-681. Delft: Nederlandse voreniging voor
Elektronenmikroscopie.
KATSURA, I. & NODA, H. (1973). Structure and polymorphism of light meromyosin aggregates.
J. Biochem. 73, 257-268.
KING, M. V. & YOUNG, M. (1970). Selective non-enzymic cleavage of the myosin rod. Electron
microscopic studies on crystals and paracrystals of light meromyosin-c. J. molec. Biol. 50,
491-507.
PHILPOTT, D. E. & SZENT-GYORGYI, A. G. (1954). The structure of light-meromyosin: an electron
microscopic study. Biochim. biophys. Acta 15, 165-173.
REIMER, L. (1965). Irradiation changes in organic and inorganic objects. Lab. Invest., Suppl. 14,
344-358.
SZENT-GY5RGYI, A. G. (1953). Meromyosins, the subunits of myosin. Arclis Biochem. Biophys.
BAHR,
42. 3O5-32O.
A. G., COHEN, C. & PHILPOTT, D. E. (i960). Light meromyosin fraction
I: A helical molecule from myosin. J. molec. Biol. 2, 133-142.
WILLIS, R. A. (1973). Optimization of stereoscopic and high angle tilting procedures for
biological thin sections. J. Microscopy 98, 379-395.
SZENT-GYORGYI,
{Received 17 December 1973)
Irradiation changes and tilted sections
B
0°
0°
Fig. 4. Electron micrographs, x 90000, of light meromyosin paracrystals. A, part of a
whole paracrystal positively stained on the grid with potassium phosphotungstate and
the excess removed with o-i M ammonium acetate, B-F, sections of embedded paracrystals some of which have been tilted about an axis perpendicular to the long axis
of the paracrystal to reveal a sharp banding pattern: B, a sharply banded paracrystal
seen without tilting; c, a paracrystal with diffuse bands at zero tilt; D, the same paracrystal as in c but tilted by 330 to show sharp-banding; E, F, sharp banding patterns seen
in other paracrystals after tilting through 45 and 180, respectively.