CHAPTER

In situ cryo-electron
tomography and
subtomogram averaging of
intraflagellar transport trains
9
Mareike A. Jordan, Gaia Pigino*
Max Planck Institute of Molecular Cell Biology and Genetics (MPI-CBG), Dresden, Germany
*Corresponding author: e-mail address: [email protected]

Chapter outline
1 Introduction......................................................................................................180
2 Methods...........................................................................................................181
2.1 Sample preparation by plunge freezing.................................................181
2.2 Data acquisition.................................................................................183
2.2.1 Electron microscope setup..............................................................183
2.2.2 Sample screening and considerations..............................................183
2.2.3 Tilt series acquisition.......................................................................183
2.2.4 Tomographic reconstruction............................................................185
2.3 Structural analysis.............................................................................185
2.3.1 Recognizing IFT trains.....................................................................185
2.3.2 IFT particle picking and pre-alignment of IFT trains..........................186
2.3.3 Subtomogram averaging..................................................................187
2.3.4 Dealing with the 6 and 11 nm repeating distance..............................189
3 Instrumentation and materials............................................................................191
3.1 Sample preparation by plunge freezing.................................................191
3.2 Electron microscope setup and data acquisition....................................191
3.3 Tomographic reconstruction and data analysis.......................................191
4 Conclusions......................................................................................................192
Acknowledgments..................................................................................................192
References............................................................................................................193

Methods in Cell Biology, Volume 152, ISSN 0091-679X, https://doi.org/10.1016/bs.mcb.2019.04.005

© 2019 Elsevier Inc. All rights reserved.
179
180 CHAPTER 9 Cryo-electron tomography of intraflagellar transport

Abstract
In situ cryo-electron tomography (cryo-ET) and subtomogram averaging are powerful tools,
able to provide 3D structures of biological samples at sub-nanometer resolution, while preserving
information about cellular context and higher-order assembly. Best results are typically achieved,
when applied to highly repetitive structures, such as viruses. Other typical examples are protein
complexes that decorate long stretches along ciliary microtubules at stereotypical and precise
repeats, such as axonemal dyneins. For such cases, a plethora of subtomogram averaging pro-
tocols exist. In this chapter, we show how we use cryo-ET and subtomogram averaging to
study the architecture of the intraflagellar transport (IFT) machinery, a more challenging target
that appears only in low copy numbers per tomogram. In the IFT trains, repeating units of IFT
adaptor proteins engage two oppositely directed molecular motors to quickly shuttle ciliary
building blocks and other proteins to the tip of the cilium and/or back to the base. This dynamic
and sporadic nature of IFT trains poses challenges for determining the localization or precise
orientation of the particles to be averaged. Solutions to these problems are described in this
chapter.

1 Introduction
The structural analysis of proteins is often coupled to in vitro conditions, though most
proteins require their natural environment to assemble and to perform their whole
function. Thanks to new developments in cryo-electron microscopy (cryo-EM), such
as direct electron detectors, it is now possible to acquire relatively highly contrasted
tomograms of thick specimens of up to 300 nm and image proteins in their natural
environment. In contrast to single-particle approaches, where only 2D images of pu-
rified proteins are acquired, tomograms allow one to slice through the volume and to
isolate the particles of interest from their surrounding cellular context in silico. The
tomographic approach, however, does not come without compromises: raw tomo-
grams are noisy and suffer from so-called missing wedge artifacts. Sample damage
due to excessive electron irradiation allows only the use of a low electron dose, which
gives rise to high noise. Missing wedge arises from the limited tilt range of the sam-
ple in the microscope, as tilts between 70° and 90° lead to excessive thickness of the
sample (Schmidt & Booth, 2008). Hence, certain spatial information cannot be
acquired and all structures are impaired in their completeness. To overcome these
problems and obtain high-resolution structures, subtomogram averaging needs to
be applied (Wan & Brigg, 2016). This includes (i) selection of numerous particles
of the same structural content, (ii) spatial alignment of all particles to one common
reference, and (iii) averaging in order to increase the signal-to-noise ratio and to fill
in the missing wedge from particles that were acquired in different orientations with
respect to the electron beam axis.
Here, we show how we use in situ cryo-electron tomography (cryo-ET) and sub-
tomogram averaging to investigate the structure and mechanism of the molecular
machinery required for the assembly and maintenance of the cilium, the intraflagellar
transport (IFT). IFT particles, which are composed of at least 22 proteins, assemble
 2 Methods 181

into trains that transport ciliary building blocks between the cell body and the ciliary
tip, moving along microtubule doublets (MTd) (reviewed in Taschner & Lorentzen,
2016). Anterograde trains are driven by the motor protein kinesin-2 (Kozminski,
Beech, & Rosenbaum, 1995). At the tip, the trains reassemble and return to the cell
as retrograde trains, powered by dynein-1b (Pazour, Wilkerson, & Witman, 1998;
Porter, Bower, Knott, Byrd, & Dentler, 1999). IFT was first discovered in the cilia
of Chlamydomonas by video-enhanced DIC microscopy (Kozminski, Johnson,
Forscher, & Rosenbaum, 1993) and it has since been extensively studied in this organ-
ism by biochemistry, DIC microscopy, and TIRF microscopy (reviewed in Taschner &
Lorentzen, 2016). Only a few IFT proteins have been described by crystallography
(Bhogaraju, Taschner, Morawetz, Basquin, & Lorentzen, 2011; Taschner, Kotsis,
Braeuer, Kuehn, & Lorentzen, 2014; Taschner & Lorentzen, 2016), while structural
studies of entire IFT trains were for a long time limited to the resolution achievable
by room-temperature EM of plastic embedded samples (Kozminski et al., 1993;
Pigino et al., 2009; Stepanek & Pigino, 2016). Therefore, the architecture of IFT trains
remained elusive until the recent publication of the cryo-EM structure of anterograde
IFT trains (Jordan, Diener, Stepanek, & Pigino, 2018).
With the cryo-ET approach described in this chapter, we were able to reconstruct
3D models of trains from the intact cilia of wild type and mutant cells. The recon-
structions provided information about the organization of the two complexes IFT-A
and IFT-B that act as adaptors between the motor and cargo proteins. The structure
furthermore illustrated how the retrograde motor dynein-1b is prevented from engag-
ing in a tug-of-war with kinesin-2 during anterograde transport (Jordan et al., 2018).
These experiments were performed in situ in the cilia of the green algae Chlamydo-
monas reinhardtii, a prominent model organism used to study the structure of cilia.

2 Methods
2.1 Sample preparation by plunge freezing
Chlamydomonas cells are cultured in TAP medium as described previously
(Gorman & Levine, 1965; Harris, 2009) for 3 days at 22 °C under a light-dark cycle
with constant aeration. Cells are used undiluted, which usually corresponded to a
sample absorbance of 0.8–1.0 at a wavelength of 750 nm.
TEM gold grids with holey carbon support film (Quantifoil Micro Tools GmbH,
Au 200 mesh R3.5/1) were glow-discharged from both sides in a plasma cleaner
(Diener electronic GmbH, Femto) to make the surface hydrophilic. 3 μL 10-nm col-
loidal gold particles (BBI Solutions) are added to the grids. After 1 min, excess liquid
is blotted away and the grids are air-dried. These gold particles are only distributed
on the carbon surface and can later be used as an additional aid for tracking and
focusing during data acquisition.
For plunge freezing, a Leica EM GP is used that provides temperature and hu-
midity control and a sensor to monitor the blotting process (Resch, Brandstetter,
K€onigsmaier, Urban, & Pickl-Herk, 2011). The temperature is set to 18 °C and
182 CHAPTER 9 Cryo-electron tomography of intraflagellar transport

humidity is held between 75% and 85%. 3 μL cells are applied to the grid and im-
mediately mixed with 1 μL 10-nm gold particles directly on the grid to avoid long
incubation of the cells with the gold. Blotting is performed from the back with What-
man filter paper for 1 s (Fig. 1A) and the grid is immediately plunged into liquid
ethane at
182 °C and stored in liquid nitrogen until data acquisition.

FIG. 1
Sample preparation and imaging setup. (A) Vitrification of Chlamydomonas cells. Whole cells
from a liquid culture are applied on a glow-discharged gold grid. The sample is blotted
from the back of the grid and rapidly plunged into liquid ethane. The process is performed in
the humidified chamber of the Leica EM Grid Plunger (GP) (A0 ). (B) Screening for
suitable cilia in the electron microscope. A grid montage (B0 ) is acquired at low magnification
of 210 . Cilia that are attached to the cell body and are approximately aligned along the tilt
axis within an ice hole can be identified. (C) Setup of a tilt series. The area of interest is
investigated at “View” magnification of 2250  and the “Record” area is defined. The “Focus
and Tracking” area is positioned on a carbon area along the tilt axis (C0 ) and far enough
from the “Record” area to avoid beam damage on the sample.
 2 Methods 183

2.2 Data acquisition

2.2.1 Electron microscope setup
For data acquisition, we use a Thermo Fisher (former FEI) Titan Halo cryo-TEM
equipped with a field emission gun, operating at 300 kV. Images are recorded on
a Gatan K2 summit direct electron detector with an energy filter (GIF, Gatan image
filter). Digital Micrograph software was used to tune the GIF and SerialEM software
(Mastronarde, 2005) was employed for the automated acquisition of tomographic tilt
series.

2.2.2 Sample screening and considerations

To identify suitable cilia for data acquisition, montages of the whole grid are ac-
quired at low magnification of 210 (Fig. 1B). These maps allow us to localize cilia
that are attached to the cell body and span across holes in the carbon film, embedded
in an ice layer. Cilia that are aligned along the tilt axis (with up to  45° tolerance) are
preferred for data acquisition. Generally, cells are not captured in gliding position,
even though they show gliding activity on the carbon film under a light microscope.
Instead, the cilia are often oriented randomly, spreading away from the cell body and
sometimes appear in parallel pairs, which are also suitable for tilt series acquisition.
Promising areas are then imaged with a magnification of 2250  to guide further
inspections (Fig. 1C). Short record images with total doses of  1 e/Å2 are acquired
with 5–10% overlap along the cilium to identify the most rewarding position of the
tomogram. IFT trains can often already be distinguished in these projection images if
they are located at the side of the cilium (Fig. 2A), and the tilt series are centered such
that as many visible trains as possible are captured. In general, biological specimen
as cilia tends to get compressed during the mechanical forces of the blotting process
(Nicastro, 2009). Especially membranated cilia with glycocalyx seem to be prone to
heavy compression, which could also result in deformation of ciliary particles and
ultimately limit resolution. The presence of IFT trains in a cilium segment increases
the thickness of the sample and thus also its susceptibility to compression deforma-
tion. In a trade-off for particle numbers, we therefore image cilia with a diameter of
up to maximal 270–280 nm (measured from membrane to membrane, without IFT),
which corresponds to slight compression but would keep the morphology of
IFT trains substantially intact. Furthermore, cilia with less than 10 well-distributed
fiducial markers are not considered for tilt series acquisition.

2.2.3 Tilt series acquisition

Tilt series are acquired using SerialEM (Mastronarde, 2005) at a magnification of
30,000 with a pixel size of 0.236 nm in super-resolution mode of the K2 camera.
The chosen magnification is a trade-off between resolution and particle number.
The captured field of view is 1.75  1.80 μm and one tomogram with one or two par-
allel cilia contains on average four IFT trains (range 0–15). The tilting scheme is
bidirectional with a starting angle of
20° and maximal tilt angles of 64° when pos-
sible. Images are acquired every 2°. The defocus range is between
4 and
6 μm,
and the slit width of the energy filter is 20 eV. Each tomogram has a cumulative dose
184 CHAPTER 9 Cryo-electron tomography of intraflagellar transport

FIG. 2
Views of anterograde IFT trains in cryo-ET with different missing wedge orientations.
(A) Projection image of a cilium, showing a bulge in the membrane that indicates an IFT train.
(A0 ) The IFT train itself can be identified after tomographic reconstruction (M, membrane;
G, glycocalyx). (B) Cryo-electron tomogram of a cilium, oriented with the tip (+) to the
right. IFT trains are indicated with arrowheads. (CP, central pair; MTd, microtubule doublet).
(C, E) Slices through an IFT train show three polymer-like structures with different repeats:
18 nm (dynein, blue), 6 nm (IFT-B, green) and 11 nm (IFT-A, yellow). IFT trains appear
in different views, depending on their orientation in the cilium with respect to the imaging
axis z and the missing wedge. (C, D) “Side view”: IFT train positioned at the side of the cilium
with respect to z. Slices through the train at different z heights show the 18 nm repeat (C) or
the 6 and 11 nm repeat (C0 ). The IFT train is easily visible in cross section since its
elongated features extend along the electron beam axis (D). (E, F) “Top view”: IFT train
positioned on top (or bottom) of a cilium with respect to z. Slices through the train at different z
heights show the 11 nm repeat (E) or the 18 nm repeat together with the 6 nm repeat (E0 ). Note
that the IFT train is barely visible in cross section (F) since its elongated features are
perpendicular to the electron beam axis. All longitudinal sections are shown with 7 nm slice
thickness, cross sections with 40 nm. (G) Cross section of a cilium to demonstrate the
effect of the missing wedge. The cilium has been imaged from
60° to +60°. (H) Illustration of
the geometry of the cilium.
Panels (B), (C) and (C 0 ) are reused from Jordan, M. A., Diener, D. R., Stepanek, L., & Pigino, G. (2018).
The cryo-EM structure of intraflagellar transport trains reveals how dynein is inactivated to ensure unidirectional
anterograde movement in cilia. Nature Cell Biology, 20(11), 1250–1255. https://doi.org/10.1038/
s41556-018-0213-1. with permission from Springer Nature.
 2 Methods 185

between 100 and 140 e/Å2 with an image dose of 1.8–2.1 e/Å2. The electron rate on the
camera is below 9–10 e/pix/s at low tilt angles to allow effective electron counting,
resulting in 3–6 e/pix/s at high tilt angles. Exposure times are between 1.6 and
2.5 s per image, while each 10 frames are acquired. The sample drift is held well
below 1 nm/s.

2.2.4 Tomographic reconstruction

In general, tomograms are reconstructed similarly as described elsewhere (Nicastro,
2009). To align the frames of each tilt image, we use the software K2Align, which is
based on the MotionCorr algorithm (Li et al., 2013). Tomogram reconstruction is
then performed with Etomo from IMOD (Kremer, Mastronarde, & McIntosh,
1996). In short, the images of a tilt series are first coarsely aligned using cross cor-
relation. For the following fine alignment, gold beads are used as fiducial markers.
To focus the alignment on the cilium, only beads in close proximity to the cilium are
selected, while beads on carbon or free ice are omitted as they often show different
moving behavior during sample irradiation. At minimum, 10 beads are selected for
each tilt series, when available 40–50 beads are used. CTF curves are estimated with
CTFPLOTTER and corrected by phase-flipping with CTFPHASEFLIP, both
implemented in IMOD (Xiong, Morphew, Schwartz, Hoenger, & Mastronarde,
2009). Dose filtration is performed to account for beam damage during the data
acquisition and remove noise above damaged frequencies (Grant & Grigorieff,
2015). To this end, the program “mtffilter” of IMOD is used, considering the
accumulative dose of each image taken to apply increasing lowpass filters to con-
secutive images. The tomograms are reconstructed by weighted back-projection
and subsequently binned by 3 and 6, resulting in pixel sizes of 0.236 nm (unbinned),
0.708 nm (bin3) and 1.413 nm (bin6). To further facilitate particle picking and
visualization, a non-linear anisotropic diffusion (NAD) filter by IMOD is applied
to the 6  binned tomograms.

2.3 Structural analysis

2.3.1 Recognizing IFT trains
Anterograde IFT trains can easily be identified in reconstructed tomograms
(Fig. 2A0 , B). For comparison with retrograde trains see Jordan et al., 2018.
It is important to notice that anterograde trains are composed of three different
repeating structures, hence they have to be averaged separately. To know how the
subunits of these repeating structures have been attributed to three of the major com-
ponents of the IFT machinery (IFT-A, IFTB, IFT-dynein) see Jordan et al., 2018.
IFT-dyneins (dynein-1b) are transported with a spacing of 18 nm (blue in Fig. 2).
Each dynein unit associates with three repeats of IFT-B, which has a tight spacing
of 6 nm (green in Fig. 2) and spans over the B-tubule. IFT-A (yellow in Fig. 2) is
positioned between IFT-B and the membrane and repeats with 11 nm spacing. As
subtomogram averaging can be applied only to particles of the same structural con-
tent, the repeats have to be picked and aligned separately. The 6 nm IFT-B can be
obtained as part of the 18 nm dynein but not of the 11 nm IFT-A. Initially, two particle
models will be created for each train, one for dynein and one for IFT-A. IFT-B will be
186 CHAPTER 9 Cryo-electron tomography of intraflagellar transport

averaged as a unit of three particles per dynein. Later, the dynein model can be inter-
spaced with a 6 nm repeat to match IFT-B, and the alignment of dynein can be used to
obtain a good initial alignment for IFT-B particles.

2.3.2 IFT particle picking and pre-alignment of IFT trains

While looking for trains in tomograms and picking particles, it has to be taken into
account that IFT trains have a significantly different appearance, depending on their
position in the cilium and orientation relative to the missing wedge (compare
Fig. 2C–F). Particle picking and subtomogram averaging is performed with the soft-
ware PEET (Heumann, Hoenger, & Mastronarde, 2011). For particle picking, the
6  binned, NAD filtered tomograms are used. The chains of IFT-dynein and
IFT-A are traced separately with relatively straight contours that can account
for slight bending of the trains (Fig. 3A and B, top panels). Model points are then

FIG. 3
IFT particle picking and pre-alignment of trains. (A, B) The train structure is approximated by
two linear models and subsequently interspaced with particle points for IFT-dynein with
18 nm distance (A) or IFT-A with 11 nm distance (B). (C) Three examples of IFT-dynein
averages from single trains. Sections through dynein (blue) and IFT-B (green) are shown.
Cross-sections show different features, depending on their orientation to the missing wedge.
Note that IFT-A (transparent yellow) is poorly resolved in these averages. Trains in
(C) and (C0 ) are referred to as “side views,” (C00 ) shows a “top view” train. (D) The individual
trains are manually pre-aligned. IFT-B serves as a guide and is oriented vertically, with
the membrane on the left and the microtubule doublet on the right. (E) Averages of IFT-A of
individual trains. Note that dynein and IFT-B are poorly resolved in these averages
(transparent blue and green).
 2 Methods 187

added at the corresponding spacing. For this purpose, PEET provides the function
“addModPts” (Fig. 3A and B, bottom panels). In order to avoid that picked positions
shift out of register with regard to actual particle positions, we adjust the picking
distance for every train individually.
In the following, all IFT trains need to be aligned to one reference. However,
global alignment searches are computationally very expensive and error-prone to
missing-wedge alignment. As IFT trains have a regular array organization and follow
the overall geometry of the axoneme, they can be pre-aligned to match a common
orientation (similar to Nicastro, 2009). To this end, averages of individual trains
are generated from dynein particles of 6  binned tomograms (Fig. 3C). Therefore,
the particles of one train are quickly aligned to one of its central particles by allowing
only very small shifts and rotations. PEET provides the function “Align particle
Y axes,” which can compensate for internal bending of one IFT train. The resulting
train averages are then manually pre-aligned, whereby the elongated structure of
IFT-B serves as helpful guide (Fig. 3D). The same rotations that were determined
for the train orientation of dynein/IFT-B are then applied to IFT-A particles of
the same train. PEET provides the function “modifyMotiveList” to apply rotations
to particles.

2.3.3 Subtomogram averaging

One of the train averages is selected as an initial reference for further subtomogram
averaging, each for dynein/IFT-B and IFT-A (Fig. 4A). Large masks are used that
contain three particles of dynein/IFT-B or five particles of IFT-A but hide the influ-
ence from the membrane or the MTd (Fig. 4A). We use the FIJI plugin Labkit
to generate such binary masks (Schindelin et al., 2012). Particles from other trains
are aligned to these references using PEET. As they are already pre-aligned
(Section 2.3.2), rotations can be kept relatively small. PEET suggests to use Y as
the axis along the IFT trains; accordingly, rotations around Y (phi) are given a larger
search range of 30° as the highest error is expected here due to manual alignment.
Theta and psi only describe the internal rotation of an IFT train and are limited to 10°.
Search ranges along X and Z are chosen according to the estimated error of picking
precision. The search along Y, in contrast, is limited to one half of the picking
distance (6 px for dynein, 4 px for IFT-A) to reduce the likelihood that particle points
would shift to neighboring particle positions. The resulting averages focus on the
train structure that was included in the mask, while the membrane and MTds become
less enhanced (Fig. 4B). The respective other IFT repeat (i.e., IFT-A in the dynein/
IFT-B average or IFT-B in the IFT-A average) are not enhanced but remain visible in
small parts. They can be used to register both averages in order to generate a whole
train model (Fig. 4E). To further improve the reconstruction of the structure, smaller
masks can be applied that contain each only one particle of IFT-B (Fig. 4C1), dynein
(Fig. 4C2) or IFT-A (Fig. 4C3).
IFT-B has so far only been obtained from the dynein average. To increase the
number of particles and optimize the average, the particle picking is repeated with
a distance of 6 nm. To this end, the particle points of dynein (18 nm) are used but
188 CHAPTER 9 Cryo-electron tomography of intraflagellar transport

FIG. 4
Subtomogram averaging strategy to resolve the three repeat patterns of anterograde IFT
trains. (A) Initial references and alignment masks. The averages from one train (also shown
in Fig. 3C0 ,D0 ,E0 ) are used as initial reference (gray), while using large masks (green and
yellow) for the alignment. Two separate particle pickings and masks are employed to resolve
for the repeat of Dynein/IFT-B (18/6 nm, (A1)) and IFT-A (11 nm, (A2)). (B) Resulting
averages of dynein (blue)/IFT-B (green) (B1) and IFT-A (yellow) (B2) after alignment.
(C) Smaller masks that contain only one particle of IFT-B (green, C1), dynein (blue, C2) or
IFT-A (yellow, C3) are applied to refine the alignment. The process is transferred from
6  to 3 binned data. (C1) The original particle picking of 18 nm from dynein is interspaced
with model points at 6 nm to meet the repeat of IFT-B and shifted from the dynein to the
IFT-B center. (D) Refined averages of IFT-B, dynein and IFT-A. (E) To construct an
entire train, both initial averages are merged by using common parts of IFT-A or IFT-B,
respectively. (F) Copies of the three refined structures are inserted into the whole train
structure to generate the final IFT train model.
 2 Methods 189

interspaced with additional points (PEET function “addModPts”). Shifts and rota-
tions of the dynein alignment can be transferred to the new IFT-B points as good
starting values, since little variability can be assumed for neighboring particles.
To this end, the entries for shifts and rotations in the PEET motive lists of the dynein
alignment are triplicated. Additionally, the center of the particle points is shifted
from the dynein center toward the IFT-B center (Fig. 4C1).
To decreases the pixel size of all three IFT units, the alignment is transferred
from 6  to 3  binned data. Therefore, the models have to be adapted with the
function "imodtrans -i" and the shifts have to be multiplied by a factor of 2. The
final averages provide refined structures of each IFT repeat unit: dynein, IFT-B
and IFT-A (Fig. 4D1–D3). To clean the structures from their surrounding environ-
ment, they are multiplied with their respective alignment masks (can be done, for
example, in IMOD with “clip multiply”). To generate a model of the whole train
structure, the two averages of dynein/IFT-B (Fig. 4B1) and IFT-A (Fig. 4B2) are
merged in UCSF Chimera (Pettersen et al., 2004). In order to position them with
respect to each other, remaining parts of IFT-B/IFT-A in the respective other
average can be used (Fig. 4E, Chimera function “Fit in map”). Finally, many copies
of the three refined structures of IFT-B, dynein and IFT-A are inserted into the den-
sity of the larger train average (Fig. 4E and F). Therefore, the averages are posi-
tioned in Chimera (Fit in map) and subsequently combined with the command
“vop maximum”.

2.3.4 Dealing with the 6 and 11 nm repeating distance

Interestingly, IFT-A and IFT-B show repeating distances (11 and 6 nm) that are not
integer multiples of each other. Direct evidence for that can be found in the power
spectra of both averages (Fig. 5A, B) and in the fact that IFT-B repeats are not fully
resolved in the average of IFT-A. One repeat of IFT-A occupies 1.83 repeats of IFT-
B, and, accordingly, their register to each other is expected to shift within the train.
In order to visualize this effect and to resolve the changing connections between
IFT-A and IFT-B, classification strategies can be applied. We use the principal
component analysis (PCA) and subsequent clustering as implemented in PEET.
In order to classify for different positions of IFT-B relative to IFT-A, a particle
mask is used that includes approximately two particles of the poorly resolved
IFT-B density within the IFT-A average (Fig. 5C). Clustering into seven classes
shows four distinct positions of IFT-B (two classes show a very similar position
and were combined; the remaining two classes contain mainly bad particles or
do not show a prominent position of IFT-B). The classes (Fig. 5D) illustrate
how the IFT-B repeat continuously shifts with respect to IFT-A. The connection
is maintained by a flexible domain between both complexes that adjusts for
the changeable distance (Fig. 5D right panels). Similar results can be obtained when
using a PCA particle mask for the poorly refined densities of IFT-A in the IFT-B
average (not shown).
190 CHAPTER 9 Cryo-electron tomography of intraflagellar transport

FIG. 5
Resolving the shift between repeats of IFT-A and IFT-B. (A) The power spectrum of IFT-B
shows a distinct layer line at a frequency of 6 nm
1, corresponding to the repeating distance of
IFT-B. (B) The power spectrum of IFT-A shows layer lines at 11 and 5.5 nm
1, which
correspond to the first and second order lines of the 11-nm repeat. Additionally, a satellite
layer line at 6 nm
1 is visible near the second order line of IFT-A and identifies the IFT-B
repeat. This shows that IFT-A has a repeating distance that is slightly smaller than two
repeats of IFT-B. (C) A particle mask (orange) that includes two repeats of the poorly resolved
IFT-B density is used for PCA of the IFT-A average. (D) Four classes show the shift of
IFT-B particles relative to IFT-A. Note that while IFT-A remains its position, IFT-B continuously
shifts toward the right (green arrows). The connecting density between both complexes
adjusts flexibly to the shift (right panel).
Panels (A) and (B) are reused from Jordan, M. A., Diener, D. R., Stepanek, L., & Pigino, G. (2018). The cryo-EM
structure of intraflagellar transport trains reveals how dynein is inactivated to ensure unidirectional anterograde
movement in cilia. Nature Cell Biology, 20(11), 1250–1255. https://doi.org/10.1038/s41556-018-0213-1.
with permission from Springer Nature.
 3 Instrumentation and materials 191

3 Instrumentation and materials

3.1 Sample preparation by plunge freezing
Instrumentation
Plasma cleaner (Diener electronic GmbH, Femto)
Plunge freezer (Leica EM GP)
Materials
Chlamydomonas reinhardtii strain cc125 (obtained from the Chlamydomonas
resource centre)
TEM gold grids with holey carbon support film (Quantifoil Micro Tools GmbH,
Au 200 mesh R3.5/1)
Whatman filter paper
Liquid nitrogen & ethane
Dewars (KGW Isotherm; Taylor-Wharton LD4)
Cryo boxes for grid storage (3D printed)
Tweezers (Dumont, L5)

Reagents
Tris-acetate-phosphate (TAP) medium (as described in Harris, 2009)
10-nm colloidal gold particles (BBI Solutions)

3.2 Electron microscope setup and data acquisition

Instrumentation

Single tilt liquid nitrogen cryo-transfer holder Model 626, Turbo pumping station
Model 655, SmartSet Cold stage controller Model 900 (Gatan)
Titan Halo 300 kV FEG cryo-TEM (Thermo Fisher, former FEI)
K2 summit direct detection camera Model 1000 & GIF Quantum LS Model 967
(Gatan)
Software

TEM user interface (Thermo Fisher)

Digital Micrograph (Gatan)
SerialEM (Mastronarde, 2005)

3.3 Tomographic reconstruction and data analysis

Instrumentation
Dell Precision Tower 7910, CentOS Linux 7
192 CHAPTER 9 Cryo-electron tomography of intraflagellar transport

Software

K2Align (provided from the Baumeister group)

IMOD (Kremer et al., 1996)
PEET (Heumann et al., 2011)
UCSF Chimera (Pettersen et al., 2004)

4 Conclusions
The cilia of Chlamydomonas are an especially rewarding target for in situ cryo-ET.
They protrude away from the thick cell body and are just thin enough to be imaged
with good contrast in a 300 kV cryo-TEM, equipped with a direct electron detector
and energy filter. The cells are directly imaged after plunge freezing, with no need for
additional time-consuming preparation steps, such as focused ion beam (FIB) milling
(Mahamid et al., 2016) or vitreous sectioning (Al-Amoudi, Norlen, & Dubochet,
2004). In this chapter we showed how we take advantage of these feature of
Chlamydomonas cilia to image IFT trains in situ by cryo-tomography and reconstruct
3D models of the major IFT protein complexes. The identification of IFT-A, IFT-B,
and IFT-dynein was achieved by mutant analysis (Jordan et al., 2018). This is a clas-
sical approach often used to locate individual components or subunits within a larger
protein complex, as knock-down or knock-out mutants result in structures with miss-
ing densities when compared to the wild type (Urbanska et al., 2015). However, the
absence of one protein often also prevents the association of other proteins. This pre-
vents the identification of its specific localization and only allows the protein to be
assigned to the density of a large sub-complex (Jordan et al., 2018; Pigino et al.,
2011). In the case of IFT proteins, complete null-mutants often result in the loss
of cilia, which prevents structural analysis. As an alternative, knock-down or
temperature-sensitive mutants have to be used (Jordan et al., 2018). A more precise
way to localize subunits is by protein tagging that results in additional densities
(Stoddard et al., 2018; Toropova, Huffman, Homa, & Conway, 2011). If the target
protein is accessible, BCCP- or SNAP-streptavidin tagging can be performed (Oda &
Kikkawa, 2013; Song et al., 2015). However, these approaches require the removal
of the membrane to make the axoneme accessible to streptavidin and hence are not
appropriate to target IFT proteins. Therefore, a combination of strategies, which
might include in situ protein tagging, will be necessary to investigate the architecture
of the IFT trains in the future.

Acknowledgments
We thank Tobias F€urstenhaupt and the EM-facility of the MPI-CBG for setting up cryo-EM in
Dresden. Dennis Diener for fruitful and continuous discussion. Oscar Gonzales from the Com-
puter Department of MPI-CBG and Anand Dwivedi for providing computational support and
scripts. The K2Align software was provided by Wolfgang Baumeister and his team. This work
was supported by the Max Planck Society.
 References 193

References
Al-Amoudi, A., Norlen, L. P., & Dubochet, J. (2004). Cryo-electron microscopy of vitreous
sections of native biological cells and tissues. Journal of Structural Biology, 148(1),
131–135. https://doi.org/10.1016/j.jsb.2004.03.010.
Bhogaraju, S., Taschner, M., Morawetz, M., Basquin, C., & Lorentzen, E. (2011). Crystal
structure of the intraflagellar transport complex 25/27. The EMBO Journal, 30(10),
1907–1918. https://doi.org/10.1038/emboj.2011.110.
Gorman, G. S., & Levine, R. P. (1965). Cytochrome f and plastocyanin: Their sequence
in the photosynthetic electron transport chain of Chlamydomonas reinhardtii. Proceed-
ings of the National Academy of Sciences of the United States of America, 54(6),
1665–1669.
Grant, T., & Grigorieff, N. (2015). Measuring the optimal exposure for single particle cryo-
EM using a 2.6 Å reconstruction of rotavirus VP6. eLife, 29(4), e06980. https://doi.org/
10.7554/eLife.06980.
Harris, E. H. (2009). The Chlamydomonas sourcebook: Introduction to Chlamydomonas and
its laboratory use. Vol. 1. Academic press.
Heumann, J. M., Hoenger, A., & Mastronarde, D. N. (2011). Clustering and variance maps for
cryo-electron tomography using wedge-masked differences. Journal of Structural
Biology, 175(3), 288–299. https://doi.org/10.1016/j.jsb.2011.05.011.
Jordan, M. A., Diener, D. R., Stepanek, L., & Pigino, G. (2018). The cryo-EM structure of
intraflagellar transport trains reveals how dynein is inactivated to ensure unidirectional
anterograde movement in cilia. Nature Cell Biology, 20(11), 1250–1255. https://doi.
org/10.1038/s41556-018-0213-1.
Kozminski, K. G., Beech, P. L., & Rosenbaum, J. L. (1995). The Chlamydomonas kinesin-like
protein FLA10 is involved in motility associated with the flagellar membrane. The Journal
of Cell Biology, 131(6 Pt. 1), 1517–1527.
Kozminski, K. G., Johnson, K. A., Forscher, P., & Rosenbaum, J. L. (1993). A motility in the
eukaryotic flagellum unrelated to flagellar beating. Proceedings of the National Academy
of Sciences of the United States of America, 90(12), 5519–5523.
Kremer, J. R., Mastronarde, D. N., & McIntosh, J. R. (1996). Computer visualization of three-
dimensional image data using IMOD. Journal of Structural Biology, 116(1), 71–76.
https://doi.org/10.1006/jsbi.1996.0013.
Li, X., Mooney, P., Zheng, S., Booth, C. R., Braunfeld, M. B., Gubbens, S., et al. (2013).
Electron counting and beam-induced motion correction enable near-atomic-resolution
single-particle cryo-EM. Nature Methods, 10(6), 584–590. https://doi.org/10.1038/
nmeth.2472.
Mahamid, J., Pfeffer, S., Schaffer, M., Villa, E., Danev, R., Cuellar, L. K., et al. (2016).
Visualizing the molecular sociology at the HeLa cell nuclear periphery. Science
(New York, N.Y.), 351(6276), 969–972. https://doi.org/10.1126/science.aad8857.
Mastronarde, D. N. (2005). Automated electron microscope tomography using robust predic-
tion of specimen movements. Journal of Structural Biology, 152(1), 36–51. https://doi.org/
10.1016/j.jsb.2005.07.007.
Nicastro, D. (2009). Cryo-electron microscope tomography to study axonemal organization.
Methods in Cell Biology, 91, 1–39. https://doi.org/10.1016/S0091-679X(08)91001-3.
Oda, T., & Kikkawa, M. (2013). Novel structural labeling method using cryo-electron tomog-
raphy and biotin-streptavidin system. Journal of Structural Biology, 183(3), 305–311.
https://doi.org/10.1016/j.jsb.2013.07.003.
194 CHAPTER 9 Cryo-electron tomography of intraflagellar transport

Pazour, G. J., Wilkerson, C. G., & Witman, G. B. (1998). A dynein light chain is essential
for the retrograde particle movement of intraflagellar transport (IFT). The Journal of Cell
Biology, 141(4), 979–992.
Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C.,
et al. (2004). UCSF chimera—A visualization system for exploratory research and anal-
ysis. Journal of Computational Chemistry, 25(13), 1605–1612. https://doi.org/10.1002/
jcc.20084.
Pigino, G., Bui, K. H., Maheshwari, A., Lupetti, P., Diener, D., & Ishikawa, T. (2011). Cryoe-
lectron tomography of radial spokes in cilia and flagella. The Journal of Cell Biology,
195(4), 673–687. https://doi.org/10.1083/jcb.201106125.
Pigino, G., Geimer, S., Lanzavecchia, S., Paccagnini, E., Cantele, F., Diener, D. R., et al.
(2009). Electron-tomographic analysis of intraflagellar transport particle trains in situ.
The Journal of Cell Biology, 187(1), 135–148. https://doi.org/10.1083/jcb.200905103.
Porter, M. E., Bower, R., Knott, J. A., Byrd, P., & Dentler, W. (1999). Cytoplasmic dynein
heavy chain 1b is required for flagellar assembly in Chlamydomonas. Molecular Biology
of the Cell, 10(3), 693–712. https://doi.org/10.1091/mbc.10.3.693.
Resch, G. P., Brandstetter, M., K€onigsmaier, L., Urban, E., & Pickl-Herk, A. M. (2011).
Immersion freezing of suspended particles and cells for cryo-electron microscopy. Cold
Spring Harbor Protocols, 2011(7), 803–814. https://doi.org/10.1101/pdb.prot5642.
Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., et al.
(2012). Fiji: An open-source platform for biological-image analysis. Nature Methods,
9(7), 676–682. https://doi.org/10.1038/nmeth.2019.
Schmidt, M. F., & Booth, C. R. (2008). Methods for aligning and for averaging 3D volumes
with missing data. Journal of Structural Biology, 161(3), 243–248. https://doi.org/
10.1016/j.jsb.2007.09.018.
Song, K., Awata, J., Tritschler, D., Bower, R., Witman, G. B., Porter, M. E., et al. (2015).
In situ localization of N and C termini of subunits of the flagellar nexin-dynein regulatory
complex (N-DRC) using SNAP tag and cryo-electron tomography. The Journal of Biolog-
ical Chemistry, 290(9), 5341–5353. https://doi.org/10.1074/jbc.M114.626556.
Stepanek, L., & Pigino, G. (2016). Microtubule doublets are double-track railways for intra-
flagellar transport trains. Science (New York, N.Y.), 352(6286), 721–724. https://doi.org/
10.1126/science.aaf4594.
Stoddard, D., Zhao, Y., Bayless, B. A., Gui, L., Louka, P., Dave, D., et al. (2018). Tetrahymena
RIB72A and RIB72B are microtubule inner proteins in the ciliary doublet microtubules.
Molecular Biology of the Cell, 29(21), 2566–2577. https://doi.org/10.1091/mbc.E18-
06-0405.
Taschner, M., Kotsis, F., Braeuer, P., Kuehn, E. W., & Lorentzen, E. (2014). Crystal structure
of IFT70/52 and IFT52/46 provide insight into intraflagellar transport B core complex
assembly. The Journal of Cell Biology, 207(2), 269–282. https://doi.org/10.1083/
jcb.201408002.
Taschner, M., & Lorentzen, E. (2016). The intraflagellar transport machinery. Cold Spring
Harbor Perspectives in Biology, 8(10), a028092. https://doi.org/10.1101/cshperspect.
a028092.
Toropova, K., Huffman, J. B., Homa, F. L., & Conway, J. F. (2011). The herpes simplex virus 1
UL17 protein is the second constituent of the capsid vertex-specific component required
for DNA packaging and retention. Journal of Virology, 85(15), 7513–7522. https://doi.org/
10.1128/JVI.00837-11.
 References 195

Urbanska, P., Song, K., Joachimiak, E., Krzemien-Ojak, L., Koprowski, P., Hennessey, T.,
et al. (2015). The CSC proteins FAP61 and FAP251 build the basal substructures of radial
spoke 3 in cilia. Molecular Biology of the Cell, 26(8), 1463–1475. https://doi.org/10.1091/
mbc.E14-11-1545.
Wan, W., & Brigg, J. A. (2016). Cryo-electron tomography and subtomogram averaging.
Methods in Enzymology, 579, 329–367. https://doi.org/10.1016/bs.mie.2016.04.014.
Xiong, Q., Morphew, M. K., Schwartz, C. L., Hoenger, A. H., & Mastronarde, D. N. (2009).
CTF determination and correction for low dose tomographic tilt series. Journal of Struc-
tural Biology, 168(3), 378–387. https://doi.org/10.1016/j.jsb.2009.08.016.