membrane protein structures without crystals, by single ... · protein structures without crystals,...
TRANSCRIPT
Membrane protein structures without crystals, by singleparticle electron cryomicroscopyKutti R Vinothkumar
Available online at www.sciencedirect.com
ScienceDirect
It is an exciting period in membrane protein structural biology
with a number of medically important protein structures
determined at a rapid pace. However, two major hurdles still
remain in the structural biology of membrane proteins. One is
the inability to obtain large amounts of protein for crystallization
and the other is the failure to get well-diffracting crystals. With
single particle electron cryomicroscopy, both these problems
can be overcome and high-resolution structures of membrane
proteins and other labile protein complexes can be obtained
with very little protein and without the need for crystals. In this
review, I highlight recent advances in electron microscopy,
detectors and software, which have allowed determination of
medium to high-resolution structures of membrane proteins
and complexes that have been difficult to study by other
structural biological techniques.
Address
Medical Research Council Laboratory of Molecular Biology, Francis
Crick Avenue, Cambridge CB2 0QH, United Kingdom
Corresponding author: Vinothkumar, Kutti R (vkumar@mrc-
lmb.cam.ac.uk)
Current Opinion in Structural Biology 2015, 33:103–114
This review comes from a themed issue on Membranes
Edited by Bruno Miroux and Eva Pebay Peyroula
For a complete overview see the Issue and the Editorial
Available online 01st October 2015
http://dx.doi.org/10.1016/j.sbi.2015.07.009
0959-440/# 2015 The Author. Published by Elsevier Ltd. This is an
open access article under the CC BY license (http://creativecom-
mons.org/licenses/by/4.0/).
IntroductionMembranes and membrane proteins have always fascinat-
ed electron microscopists because they were prominent
features in the early electron micrographs of the cells. In
the early days, electron micrographs were used to infer the
organization of the lipid and protein components and for
postulation of the fluid-mosaic model of membrane struc-
ture [1]. Specialized membranes with abundant proteins
were also identified by electron microscopy, often with the
constituent proteins arranged in crystalline arrays [2,3].
This paved way for a separate discipline within electron
microscopy, which was later called electron crystallogra-
phy. The pioneering work of Henderson and Unwin on
glucose embedded two-dimensional (2D) crystals of bacte-
riorhodopsin (also called purple membrane) revealed the
first architecture of an integral membrane protein [4,5].
www.sciencedirect.com
Subsequent development of the rapid freezing of speci-
mens in thin aqueous films gave rise to the field of electron
cryomicroscopy (cryoEM) [6��,7]. In combination with the
introduction of field emission guns and better vacuums in
the electron microscopes, this led to a great expansion of
cryoEM, including an atomic model of bacteriorhodopsin
[8,9]. It was realized at the same time that membrane
proteins in presence of lipids can also form helical tubes,
which by helical reconstruction can also provide high-
resolution maps as shown for the nicotinic acetylcholine
receptor (AChR) [10]. Thus, in the 90s when obtaining
well-diffracting three-dimensional (3D) crystals of mem-
brane proteins for X-ray crystallography was still difficult,
electron microscopy of membrane proteins in the form of
2D crystals or helical tubes was seen as an alternative
approach to obtain structures of membrane proteins.
The added advantages of the native environment and
the ability to study conformational changes as shown for
bacteriorhodopsin and AChR [11,12] prompted many
researchers to pursue electron crystallography. However,
it soon became clear that obtaining well-ordered 2D crys-
tals or tubes were just as difficult as obtaining well-ordered
3D crystals. Although many membrane proteins formed
2D crystals immediately, often they were poorly ordered
resulting in medium to low-resolution maps, with only a
handful of structures determined to high-resolution [13].
With the advent of genome sequence determination and
identification of homologues of many important membrane
proteins, there has been a rapid increase in structures of
membrane proteins mainly by 3D crystallization and X-ray
crystallography, superseding electron crystallography [14].
Obtaining sufficient membrane protein and well-diffract-
ing crystals still remains an obstacle, and having worked
on membrane proteins for a while, I have always won-
dered if it is possible to completely get away from crystals.
Two techniques that have the potential to provide high-
resolution structures of membrane proteins (or any mac-
romolecule) without crystals include nuclear magnetic
resonance spectroscopy (NMR) and single particle
cryoEM. Of these, structural determination of membrane
proteins by NMR requires a large amount of labeled
protein and stability in a given detergent for at least
few days during measurement [15]. For many membrane
proteins, in particular those from eukaryotes this may not
be feasible. In single particle cryoEM, membrane pro-
teins in solution are rapidly frozen and imaged with
electrons as single molecules in a thin film of buffer.
By averaging a large number of particles, high-resolution
structures can then be obtained with very little protein.
Current Opinion in Structural Biology 2015, 33:103–114
104 Membranes
This technique has an added advantage that distinct
structural states in a solution can be computationally
separated and multiple structures of the same protein
determined.
Single particle EMElectron micrographs of single protein molecules contain
projections of the underlying structure of the specimen and
due to the limited electron dose required to minimize
radiation damage have a poor signal to noise ratio. Averag-
ing many projections increases the signal and from these it
is possible to generate a three-dimensional 3D reconstruc-
tion (see Figure 1 for examples of electron micrographs and
Figure 1
(a)
Complex I (cymal-7)
C
γ-secretase
γ
R
(b)
Membrane protein imaging as single particles by electron cryomicroscopy.
detergent (Complex I), amphipol (g-secretase) and nanodisc (ryanodine rece
Falcon II detector and the g-secretase was imaged with the Gatan K2 Sum
fluorinated octyl maltoside was added to the protein solution just before fre
seen in these images is due to the use of relatively high dose and sufficient
(72 frames) at a total dose of �68 electrons/A2. This high dose image was o
For subsequent refinements the last 40 frames were discarded [58�]. (b) Re
ryanodine receptor. The panel shows a selection of 2D class averages of di
around the protein (visible as a band around the membrane part and marke
domain (marked with red arrow) and the quaternary structure. In the case o
be seen in the class averages. The box sizes in the 2D class averages are 2
pixel) and ryanodine receptor (1.45 A/pixel) respectively.
Current Opinion in Structural Biology 2015, 33:103–114
class averages). The basic computational principle in single
particle cryoEM involves accurate determination of five
parameters: namely the three Euler angles and the x,ytranslations. A simple rule suggests that the more accurate
the orientation of the particles, the higher the resolution.
Often, this depends on various parameters including the
signal to noise ratio, orientation distribution of the particles
that is if all possible views of the molecule are observed,
and structural heterogeneity. One or all these parameters
affect the achievable resolution. From theoretical estimate,
Henderson has calculated that if perfect images of macro-
molecules can be obtained from an electron microscope
then high-resolution structures can be reconstructed for a
omplex I
(amphipol)
-secretase
Ryanodine receptor (nanodiscs)
yanodine receptor
Current Opinion in Structural Biology
(a) Selected areas of micrographs of membrane proteins observed in
ptor). The images of Complex I and RyR were taken with the FEI
mit detector. Scale bar is 400 nm. In the case of RyR receptor,
ezing to get better distribution of the receptor [38�]. The high contrast
defocus. For example, the Complex I images were captured for 4 s
nly used for particle picking and contrast transfer function estimation.
ference-free 2D class averages of Complex I, g-secretase and
fferent membrane proteins revealing the prominent detergent/lipid belt
d with white arrow), transmembrane (TM) helices in the membrane
f Complex I, the location of some of the supernumerary subunits can
80, 90 and 384 pixels for Complex I (1.72 A/pixel), g-secretase (2.8 A/
www.sciencedirect.com
Membrane protein structures without crystals Vinothkumar 105
wide range of macromolecules by averaging only few
thousands of asymmetric units [16��,17�]. He pointed
out that radiation damage, beam-induced movement and
signal to noise ratio in images were the major limiting
experimental factors, which when solved would allow to
obtain perfect images and higher resolutions as predicted
from theory.
The advent of microscopes with coherent beams, better
vacuums and optics allowed the first atomic resolution
maps to be obtained by single particle reconstruction of
highly symmetrical macromolecules with images taken on
film or a charged-coupled device (CCD) [reviewed in 18].
These maps showed that it was indeed possible to obtain
high-resolution structures without crystals. In the last
couple of years, with the introduction of direct electron
detectors, it is now possible to pursue structural studies of
less symmetrical macromolecules that were otherwise
difficult to determine. The advantages of the direct
electron detectors can be summarized in the following
points: 1) higher detective quantum efficiency (DQE) or
simply higher signal to noise ratio [19]; 2) the possibility
to look and evaluate the quality of the specimen imme-
diately (unlike films, which had to be developed and
checked by optical diffraction) and collect more data in a
given period of time; 3) as they operate in rolling shutter
mode, the ability to dose-fractionate the exposure allows
one to computationally correct for drift and beam-induced
motion and choose selected frames and electron dose that
gives the best reconstructions [20–24]. There are several
reviews that chart the development of cryoEM from its
early history to the present day advances [25–30,31��],which I highly recommend to readers but this review will
mainly focus on single particle studies of membrane
proteins.
In the early days, electron microscopy of membrane
proteins as single particles was largely limited to visual-
izing them with heavy metal staining. This was based on
the general thought that due to the presence of detergents
it would be difficult to get sufficient contrast in ice.
However, studies on the mitochondrial Complex I [32],
hexameric P-type H+-ATPase from Neurospora crassa [33]
and the bovine F1Fo ATPase [34] showed that single
particle imaging of membrane proteins in ice was possi-
ble. Despite their limited resolution, the calculated maps
were informative and reflected the true structure of
protein (not the envelope from the heavy metal stain).
The structure of bovine ATPase in particular remains a
major landmark as it provided an envelope for an intact
multi-subunit membrane protein complex on which crys-
tal structures of several known subunits could be docked
[35].
Seeing is believing — biochemistry is vitalA great advantage of EM is the ability to directly visualize
the protein of interest and tune the biochemistry to
www.sciencedirect.com
optimize the specimen for any given investigation. The
homogeneity of membrane proteins is generally judged
based on gels and gel-filtration profiles, occasionally com-
plemented with functional assays. However even a single
band on a gel or a symmetrical gel filtration profile may
not reflect the true nature of membrane protein in solu-
tion. When observing such solutions by cryoEM, one may
sometimes be surprised to see a wide variety of features
including lipid-detergent mixtures either as vesicles or
small membranes, denatured protein or dissociated sub-
units. If the membrane protein of interest is spherical
then it can be difficult to judge the quality of biochemis-
try amid such background. With multi-subunit membrane
protein complexes, the dissociation of subunits can some-
times occur during freezing of protein solution on grids
and some understanding behind the process of blotting
and freezing is thus beneficial [6��]. In the initial stages of
studying a membrane protein by single particle cryoEM,
it is often useful to image them with a combination of high
dose (when using direct detectors) and high defocus, so
that they are clearly visible and some features of the
protein can be recognized. This is particularly useful for
smaller membrane proteins. Such high-contrast low noise
images provide a wealth of information about the purity
and integrity of the sample but also prevent picking
features that could be just low-contrast random noise [36].
Just as in crystallization, the choice of detergent or its
substitutes and the final detergent concentration plays a
critical role in cryoEM. Too high a concentration of
detergent adds to the background. When possible it is
ideal to use the protein directly off the column without a
concentration step. Membrane proteins have been ob-
served as single particles in detergent micelles, amphipol,
liposomes and nanodisc (Table 1 and Figure 1a). To date,
except for liposomes, all other media have produced
medium to high-resolution structures so at present it is
not possible to generalize about whether a particular
detergent or its substitute is best suited for cryoEM or
is essential to obtain high-resolution. The recent struc-
tures of ryanodine receptor (RyR) in detergent [37�],nanodiscs [38�] or mixed lipid-detergent micelles [39�]reveal similar structures and highlight the fact that the
specimen preparation is likely to play a more crucial role
(e.g. thickness of ice) rather than the choice of detergent
or its substitute. For a given membrane protein, it is
important to find the right mixture of detergent/lipid that
keeps in its functional and native state rather than fol-
lowing a particular protocol that has been found to be
successful for other proteins.
Reference free 2D class averaging, where particles with
similar views are averaged to produce projections of the
structure that are less noisy and can further provide assess-
ment of the quality of the protein prep. This requires only a
few thousand particles and perhaps 50–100 images. In
Figure 1b, examples of 2D class averages of different
Current Opinion in Structural Biology 2015, 33:103–114
106 Membranes
Table 1
Compilation of membrane protein structures determined by single particle electron cryomicroscopy to 10 A or below (till April 2015)
Protein Source Molecular
mass
(in MDa)
Medium Detector No. of
particlesaNo. of
asymmetric
unitsa
Resolution
(A)bEMDB Ref
Complex I Native (bovine) 1 Cymal-7 Falcon II 25,492 25,492 �5 2676 [58�]
V/A-ATPase Native (Thermus
thermophilus)
0.6 DDM Film 46,105 46,105 9 5335 [47�]
Native (Manduca
sexta)c0.9 C12E10 Falcon II 6714 6714 9.4 2781 [61�]
Native (S.cerevisiae) 0.9 DDM K2 50,030 50,030 6.9d 6284 [64�]
F1Fo-ATPase Native
(Polytomella sp.)
1.6 DDM Falcon II 37,238 74,476 �7 2852 [60�]
Glutamate
receptorcRecombinant 0.4 DDM Falcon II 21,360 42,720 7.6 2685 [57�]
Ryanodine
receptorcNative (rabbit
skeletal muscle)
2.2 Nanodisc Falcon II 25,000 100,000 6.1 2751 [38�]
Nanodisc TVIPS F816 94,354 377,416 8.5 2752 [38�]
CHAPS/lipids K2 46,400 185,600 4.8 6106 [39�]
Tween-20 Falcon II 65,872 263,488 3.8 2807 [37�]
DDM CCD 28,036 112,144 9.5 1275 [45]
CHAPS Film 25,722 102,888 10.2 5014 [46]
TRPV1 Recombinant 0.3 Amphipol A8-35 K2 37,310 149,240 3.3d 5778 [55��]
TRPA1 Recombinant 0.7 PMAL-C8 K2 20,733 82,932 �4d 6268 [63�]
g-Secretase Recombinant 0.17 Amphipol A8-35 K2 144,545 144,545 4.5 2677 [56�]
Digitonin K2 177,207 177,207 4.3 2974 [65]
Tmr AB+ AH5 Recombinant 0.18 DDM K2 102,000 102,000 8.2 6085 [59�]
Tmr AB Recombinant 0.135 DDM TVIPS F816 36,000 36,000 10 6087 [59�]
Anthrax prepore
toxin
Recombinant 0.44 NP-40 K2 60,455 423,185 2.9 6224 [62�]
Ribosome
complexese2.6–4.3
Sec61 Native (porcine) Digitonin Falcon II 80,019 80,019 3.35–3.9 2644, 46,
49, 50
[66]
Native (Canis sp.) Digitonin TVIPS F416 162,655 162,655 6.9 2510 [50]
Ssh1 Native (yeast), Ssh1(R) Digitonin Film 183,000 183,000 6.1 1651 [48]
SecYEG Native (E. coli) Nanodisc Film 85,664 85,664 7.1 1858 [49]
SecYEG (R)
SecYEb Native (M. jannaschii) DDM Film 37,000 37,000 9 5691 [53]
SecYEb (R)
Sec61+OST+
TRAP
Native (Triticum
or Canis sp.)
Digitonin TVIPS F416 15,705 15,705 9.3 2523 [52]
Digitonin Film 79,000 79,000 8.7 1528 [51]
YidC Native (E. coli)
YidC (R)
DDM TVIPS F416 58,960 58,960 8.0 2705 [54]
a Number of particles denotes the number used in the final map and number of asymmetric units is the total number averaged after the application of
symmetry.b The criteria used to estimate the resolution is the Fourier shell correlation (FSC) either at FSC at 0.143 or 0.5. The numbers denoted in bold are
reported at FSC 0.5.c The structure of V-ATPase from Manduca sexta, glutamate receptor and the early structures of ryanodine receptor (taken on film or CCD) have an
overall resolution of 10 A or better. However, the membrane domain is of lower resolutions and TM helices are not clearly resolved.d Multiple maps of V-type ATPase, TrpV1 and TrpA1 have been deposited and only the highest resolution is shown in the table. Three conformational
states of V-ATPase have been resolved and they are deposited in EMDB codes 6284–86. The structures of TrpV1 with ligands are deposited in EMDB
codes 5776–77 and please refer to Cao et al. 2013 for details of the structures [95�]. Similarly, the structures of TrpA1 with other ligands are deposited
in EMDB codes 6267 and 6269.e A collection of membrane proteins in complex with ribosomes are listed here. There are many more maps deposited in the database describing
different states and the list here is only a small collection chosen based on the species and describing only the highest resolution map. The resolution
shown is the overall resolution and in many of these maps, the membrane protein is of lower resolution, sometimes the TM-helices in the membrane
domain is not clearly resolved. The translocon proteins SSh1, SecYEG (E. coli) and SecYE (M. jannaschii) and YidC are all recombinant (shown in
brackets as ‘R’), while the ribosomes used in the study are native.
Abbreviations used in the table: EMDB: electron microscopy data base; DDM: n-dodecyl-b-maltopyranoside; Cymal-7: 7-cyclohexyl-1-heptyl-b-
maltoside; PMAL-C8: poly(maleic anhydride-alt-1-decene) substituted with 3-(dimethylamino) propylamine; C12E10: polyoxyethylene(10) dodecy-
lether; NP-40: Nonidet P-40; OST: oligosaccharyl-transferase; TRPV1: transient receptor potential V1; TRPA1: known for its extensive ankyrin
repeats at the amino terminal domain; TRAP: translocon associated protein complex.
Current Opinion in Structural Biology 2015, 33:103–114 www.sciencedirect.com
Membrane protein structures without crystals Vinothkumar 107
membrane proteins are shown. They clearly reveal distinct
views, internal details such as the transmembrane (TM)
helices and the detergent/lipid belt around the membrane
domain. The quaternary structure, integrity and conforma-
tional heterogeneity of the proteins become prominent in
such class averages (e.g. the tetrameric arrangement of RyR
in Figure 1b). Analysis from specimen preparation to
imaging and 2D classification can be done in a relatively
short period and provides a good feedback about biochem-
istry before large-scale data collection and reconstruction is
attempted. Such screening of membrane proteins and their
complexes, either by negative staining or by cryoEM, has
been instrumental in identifying the right constructs, lead-
ing to better crystals and structures and to study underlying
structural heterogeneity in solution [40–42].
From blobs to high-resolution structuresIn the early days of cryoEM of membrane proteins, a large
number of reconstructions have been published but they
showed relatively featureless blobs. There was also the
separate problem that reconstructions of the same protein
often yielded different results and interpretation. Fur-
ther, the propensity to average and refine noise resulted in
claims of resolution that could not be justified by the
maps [43,44]. Therefore, there was general skepticism
about whether reliable, high-resolution maps could be
obtained by single particle cryoEM. The maps of the
ryanodine receptor [45,46] (the membrane domain in
these RyR maps were of lower resolution) and the A
type-ATPase from Thermus thermophilus reconstructed
from images taken on a CCD or film were the first
examples to show that sub-nanometer resolution of mem-
brane proteins by cryoEM was possible [47�]. Multiple
structures of membrane proteins (the translocon, YidC) in
complex with ribosomes have been described but the
resolutions of the membrane protein part in many of these
reconstructions are generally lower than ribosomes and
often the TM helices are not clearly resolved [48–54].
Nevertheless, they have provided important biological
insights into the mechanism of protein translocation.
With direct electron detectors, it is now possible to
routinely visualize membrane proteins and obtain medi-
um to high-resolution maps in a relatively short period
(Table 1 and Figures 2 and 3). Some of the recent
structures of membrane proteins determined by single
particle cryoEM include transporters, enzymes, ion chan-
nels and multi-subunit membrane protein complexes
[37�,38�,39�,55��,56�,57�,58�,59�,60�,61�,62�,63�,64�,65].
The molecular mass of these proteins varies between
0.13 and 2.2 MDa. They contain 12–78 TM helices and
many of them lack any symmetry making these structures
truly ground breaking. The structures of RyR and ribo-
some in complex with Sec61 clearly highlight the advan-
tage of the direct detectors over film or CCD with an
increase in the resolution of the reconstructions despite
using fewer particles [37�,38�,39�,66].
www.sciencedirect.com
Many membrane proteins by nature are dynamic and can
exhibit multiple conformations in solution. For crystalli-
zation, such structural heterogeneity has been overcome
by stabilizing a particular conformation using mutagene-
sis, antibodies or inhibitors. In cryoEM, such structural
heterogeneity can be computationally sorted into distinct
classes using recently introduced maximum likelihood
based 3D classification, which results in maps that are less
noisy and far superior [67,68]. Many of the membrane
protein structures listed in Table 1 have greatly benefit-
ted from such 3D classification and I foresee that this is
one area in image processing where future improvements
in software will help immensely in obtaining higher-
resolution maps of membrane proteins in distinct con-
formations with less particles.
One of the key areas in cryoEM where there has been lot
of ambiguity is the correctness and the actual resolution of
the maps [69]. In the past few years, several tests have
been introduced to validate cryoEM maps. Of these, the
tilt-pair validation test is a very powerful technique,
where two images are taken from the same area at
different tilt angles; this then allows the orientation of
a given particle to be checked against a map that is
thought to represent the protein of interest. The hand-
edness of the map can also be determined with this
technique [17�,70]. CryoEM images are noisy and invari-
ably noise builds up during the 3D refinement. High-
resolution noise substitution or randomization of phases
beyond certain resolution is an elegant way to check over-
fitting of the data [71]. Recently introduced ‘gold-stan-
dard refinement’ aims at preventing such over-fitting
already during refinement [72,73]. As the data quality
has improved, the maps have become better and it has
become now standard to expect certain structural features
to be seen with a given claim of resolution. For instance,
maps at resolutions better than 10 A should clearly show
the separation of a-helices. Resolving b-strands at �4.5 A
and the appearance of side chain densities of larger amino
acids at resolutions beyond 4 A are the next landmarks in
judging the resolution of the maps. It is also becoming
clear that the resolution often varies across the macro-
molecule with the core regions of the protein generally
show higher resolution than the peripheral regions of the
molecule. It is now common in cryoEM publications to
show a figure of the map depicting the local resolution
[74]. The gallery of membrane protein structures in
Figures 2 and 3 are in the resolution range of 3.3–9 A
and illustrate the structural features that one would
expect for each of these resolutions. As the resolutions
have started to reach 3 A or above, bound ions and ligands
are now being built confidently. Such model building and
refinement on cryoEM maps have benefitted from mod-
ifications of existing crystallographic tools [75–78].
For a biochemically well-characterized reasonably sized
membrane protein, with the present technology it is now
Current Opinion in Structural Biology 2015, 33:103–114
108 Membranes
Figure 2
(Nicastrinectodomain)
(A1)
(A0)
IN
OUT
γ-secretase(20 TM, 0.17 MDa, 4.5 Å)
TRPV1(24 TM, 0.3 MDa, 3.3Å)
A-ATPase(~32 TM, 0.6 MDa, 9Å)
(Fab)
ABC transporter(12 TM, 0.13 MDa, 8.2Å)
Current Opinion in Structural Biology
A gallery of membrane protein structures determined by single particle cryoEM. The maps have been selected to show a range of molecular
weight, complexity of the proteins and resolution. The heterodimeric ATP-binding cassette (ABC) transporter from Thermus thermophilus has
12 TM helices with an extended cytoplasmic ATP binding domain and has a molecular mass of �130 kDa. Such a small protein by itself can be
visualized and a �10 A map has been obtained [59�]. A complex of this ABC transporter with an Fab antibody fragment (50 kDa, shown in red)
that recognizes one of the subunit aids in more accurate orientation resulting in an 8 A map that clearly resolves the TM helices. g-Secretase is an
intramembrane aspartyl protease that cleaves a wide range of single pass TM substrates but widely known for cleaving amyloid b-peptide. Four
different subunits (PS, PEN-2, APH-1 and Nicastrin) are essential for assembly and functionality of the enzyme. The protein mass of g-secretase is
�170 kDa and the presence of the large ectodomain of Nicastrin greatly assists the alignment of the particles and a map at overall resolution of
4.5 A was obtained. The ectodomain is at higher resolution and much of the path of the polypeptide could be traced [56�]. The TM domain of g-
secretase is at lower resolution but sufficient to reveal the arrangement of TM helices, providing insights into how the substrate might approach
the interior of the enzyme. The transient receptor potential V1 (TRPV1) ion channel is a receptor for capsaicin and belongs to the family of ion
channels that are involved in sensing and transducing temperature [55��]. The structure of TRPV1 at 3.3 A is one of the highest resolution maps of
a membrane protein determined by single particle cryoEM. The map shows the classical fold of the voltage-gated ion channels with 6TM helices
per monomer in a tetrameric arrangement. Such a high-resolution map allowed de novo model building of almost the entire polypeptide chain
[55��]. Multiple structures of TRPV1 in complex with inhibitors have been obtained providing insights into the possible gating mechanism and
illustrate the power of cryoEM [95�]. The H+-driven A-type ATPase from Thermus thermophilus is a moderate sized membrane protein and the
reconstruction was performed with images captured on film. Though of lower resolution (�9 A), the structure shows the global architecture of the
A type-ATPases with the A1 catalytic domain, the Ao membrane domain, the central and the two peripheral stalks [47�]. The F and V-type ATPases
have equivalent F1/V1 and Fo/Vo domains that perform catalysis and ion translocation [60�,64�]. The horizontal helices in the Ao membrane domain
that we now know to exist from higher resolution maps of the F and V-type ATPases [60�,64�] can now be correlated in this low-resolution map.
routinely possible to get maps below 10 A with very little
effort. Depending on size, symmetry and sample hetero-
geneity, achieving higher resolutions may take a substan-
tial amount of time. The overall B-factor of the map, which
reflects the quality of the images and orientation accuracy,
is a useful indicator on how well the refinement is progres-
sing and if the imaging conditions are optimal. In particular,
it provides an estimate to the additional number of particles
required to reach a particular resolution [17�]. Thus, it is
advisable to set intermediate goals and it may be useful to
improve the biochemistry, specimen preparation and im-
aging rather than averaging millions of particles to find that
Current Opinion in Structural Biology 2015, 33:103–114
the resolution of the reconstruction increases only margin-
ally. Even at medium resolutions (such as 5–7 A) biological
insights can be inferred. Such examples from recent
structures include the description of the arrangement of
TM helices and the position of the ectodomain in g-
secretase, the identification of horizontal TM helices in
ATPase’s and the assignment and location of subunits in
bovine mitochondrial Complex I (Figures 2 and 3)
[56�,58�,60�,64�]. In the end resolution is just a number.
If a biological question can be addressed with a medium
resolution map then perhaps time can be better spent on
other problems.
www.sciencedirect.com
Membrane protein structures without crystals Vinothkumar 109
Figure 3
B14.7
B16.6PGIV
(b)
(13 kDa)
Inner membrane
Matrix
(18 kDa)
42 kDa
B8
SDAP-α
SDAP-β
(MWFE+ B9)
B14(B13)
(a)
(c)
Current Opinion in Structural Biology
Architecture of mammalian respiratory Complex I. In the inner membrane of mitochondria (and the inner membrane of prokaryotes), the first step
in electron transfer is the oxidation of NADH by a multi subunit complex called NADH:Ubiqunione oxidoreductase or commonly called Complex I.
The electrons from NADH are transferred to ubiquinone through a series of iron-sulfur (FES) clusters and this transfer is coupled to proton
translocation across the membrane. The core subunits as defined in prokaryotes harbor all the catalytic subunits for electron transfer and proton
translocation [96]. In higher eukaryotes, the core subunits are augmented with varying numbers of accessory subunits that are involved in
assembly and regulation and are called supernumerary subunits. The bovine Complex I is a biochemically well studied enzyme with a molecular
mass of �1 MDa and is composed of >40 subunits [58�]. (a) A reconstruction of bovine Complex I at �5 A was obtained by cryoEM imaging of
the enzyme in detergent micelles. Two maps are overlaid to show two distinct feature of the enzyme. The eight FES clusters (shown in red) are
visible as the highest peaks in the map and the density for protein (gray) at intermediate threshold shows a number of TM helices spanning across
the membrane domain. For clarity, the detergent/lipid belt that is visible at lower threshold is not shown but the black bars mark the boundary
defined by them. (b) Assignment of subunits to Complex I. The map of Complex I has a nominal resolution of �5 A and some of the core regions
in particular TM helices are better resolved and start revealing the bulky side chains. The 14 core of subunits of bovine Complex I share high
homology with the prokaryotes. Using the atomic model of the Thermus thermophilus enzyme, the backbone and well resolved regions were built
[96]. For the assignment of supernumerary accessory subunits, a combination of biochemical and genetic information, secondary structure
prediction and subunits with known structures were employed. The identities of 14 supernumerary subunits have been assigned. These include
subunits that are membrane embedded as well as soluble proteins. The assigned subunits are colored in red and green and labeled with the
same color. The identities of the subunits whose names are labeled with brackets are less certain. Structural elements including several of the
supernumerary membrane subunits with a single TM helix have not been assigned (blue) in the current map. Figure reproduced with permission
from [58�]. (c) A view from matrix of mitochondria showing the arrangement of the TM helices in the membrane domain. The seven core
membrane subunit TM helices are shown in blue and the TM helices of supernumerary subunits in red. In total, bovine Complex I has 78 TM
helices. The curved nature of the membrane domain and the long horizontal helix are clearly visible. The detergent/lipid belt observed in the
cryoEM map is colored in green.
Present limits and possibilities of singleparticle cryoEMA question that is often asked is the lower limit in the size
of the protein for which a structure can be obtained by
www.sciencedirect.com
single particle cryoEM. According to theory, structures of
proteins as small as hemoglobin (�64 kDa) can be
obtained [16��]. While such small proteins can be visual-
ized in micrographs, they cannot be oriented reliably with
Current Opinion in Structural Biology 2015, 33:103–114
110 Membranes
Figure 4
>10º
low
0.2º
high Orientation accuracy
Large
Viruses
Ribosomes(2-4 MDa)
ComplexI(~1 MDa)
(0.4 MDa )
γsec(0.17 MDa)
ABC(0.13 MDa)
TRPV1(0.3 MDa)
β-gal
(>2 MDa)
SmallMolecular weight (MDa)
0.5 Difficulty 10
BSA(0.07 MDa)
Current Opinion in Structural Biology
Structure determination of macromolecules by single particle electron
cryomicroscopy: present limits and possibilities. With the present
technology that includes microscopes with better vacuum, stable
stage, coherent beam, direct electron detector, software and
computing we can now achieve high-resolutions of macromolecules
by single particle cryoEM. Although, the theory says that structure
determination of macromolecules can be done with less particles and
smaller proteins, currently there are limits to what is possible [16��].
As explained in the introduction of the main text, the quality of the
map/reconstruction for a macromolecule depends on how accurately
it can be oriented. Large symmetrical molecules such as viruses can
be oriented accurately between 0.2 and 0.5 degrees and maps below
4 A can now routinely be obtained with around 1000 particles (due to
symmetry the number of asymmetric units averaged will be 60,000).
Ribosomes with their bound RNA have higher contrast and are less
sensitive to radiation damage. They have been one of the test
specimens in the development of single particle EM. The resolution of
ribosome maps has been gradually increasing and multiple
reconstructions with many different factors now dominate the
electron microscopy data bank (EMDB). Thus, in terms of difficulty
high-resolution structures of large symmetrical molecules and high
contrast objects such as ribosomes can be obtained with relative
ease. As the size of the macromolecule becomes smaller, it is
generally more difficult to obtain high-resolution structures and
requires a lot more effort. Small protein molecules such as
hemoglobin or bovine serum albumin (BSA) can be visualized in
micrographs but presently cannot be oriented accurately. Recent
structures of g-secretase (170 kDa) and ABC transporter
(130 kDa + 50 kDa Fab fragment) are highlighted in red to show the
smallest asymmetric structures determined to sub-nanometer
resolution by cryoEM at the moment [56�,59�]. The number of
particles required for a given protein to reach a resolution beyond 4 A
Current Opinion in Structural Biology 2015, 33:103–114
the present technology. Figure 4 tries to summarize the
current possibilities with cryoEM for macromolecules in
general. Large symmetric molecules such as viruses can
be oriented with high accuracy and at present a resolution
of 4 A or above can be obtained by averaging only few
thousand particles requiring less than half day of data
collection. As the size of the molecule decreases the
difficulty in obtaining such high-resolution structures
increases and often require averaging of 104–106 of asym-
metric units. Dynamic and structurally heterogeneous
membrane proteins such as the rotary ATPase’s may
require much more particles to classify distinct conforma-
tions and may still only give lower resolution maps
[60�,64�].
The structure of the 130 kDa heterodimeric ATP-bind-
ing cassette (ABC) transporter from T. thermophilus is to
date the smallest asymmetric protein structure that has
been determined to below 10 A by single particle cryoEM
[59�]. This was possible by the addition of a 50 kDa Fab
fragment, which binds to one of the monomer thus not
only increasing the size of molecule but greatly aiding the
orientation determination [59�]. Similarly, in the multi
subunit complex of 170 kDa g-secretase, the presence of
a large ectodomain helps in alignment of the molecule
[56�]. For small membrane proteins that may look like
spherical blobs (in particular if the membrane part is
covered with a large amount of detergent) it is possible
that addition of Fab fragment or small domains may
become a generic technique to obtain accurate orienta-
tion [79]. These examples represent what is currently
possible with cryoEM and highlight that not only the size
but also the shape matters for obtaining high-resolution
maps by single particle EM.
Conclusions and outlookRecent progress in the field of cryoEM has been aptly
called a ‘resolution revolution’ [80]. Some of the recent
examples show that it is possible to reach beyond 3 A
[62�,81–83]. While such resolutions are remarkable, it
does involve averaging a large number of particles, in
the range of 105–106 asymmetric units (Table 1). From
theory it is clear that similar resolutions should be
obtained with fewer particles [16��]. Analysis of the movie
frames from direct electron detectors reveals that the first
few frames are worse and have much less information
than expected [24,84]. This is one of the major reasons
why the images are still not perfect and why the averaging
of so many particles is still required. In current refinement
strategies, this is overcome either by discarding those
will depend on various factors but will largely be determined by its
stability and heterogeneity and one could expect to average 104–106
asymmetric units to achieve a resolution that can resolve side chains
for a wide range of molecules.
www.sciencedirect.com
Membrane protein structures without crystals Vinothkumar 111
frames [23,82] or by down-weighting them [24,84]. Look-
ing ahead, the next technical advances in cryoEM will
come from two fronts. Firstly, understanding and curing
the problems associated with the initial frames and sec-
ondly, the next generation of direct electron detectors
with even higher DQE.
With respect to membrane proteins, won’t it be wonder-
ful if it becomes possible to visualize and obtain struc-
tures of membrane proteins in situ in their native
environment? For instance, ion channels at the synapse.
The technique of electron cryotomography, which has
been developed for a very long time, has yielded stunning
images of whole cells and thin sections [85]. With the
direct electron detectors and sub-tomogram averaging, it
is now possible to obtain structures of macromolecules in
their native environment, sometimes to sub-nanometer
resolution [86,87]. Recent analysis of the ATPase dimers
in the inner mitochondrial membrane, membrane associ-
ated mitochondrial ribosomes and the nicotinic acetyl-
choline receptor with rapsyn on the post-synaptic
membranes of the electric organ of Torpedo highlight
such possibilities for membrane proteins [88–90]. Two
recent advances namely the gold supported grids that
minimize some beam-induced movement [91�,92�] and
the phase plate [93,94] have great potential in acquiring
better images by tomography and increased resolution by
sub-tomogram averaging.
The future of cryoEM looks very promising and once a
few technical problems are overcome, in the near future it
may be possible to collect multiple data sets in a single
day just like the present day data collection from crystals
at synchrotrons.
Conflict of interestNothing declared.
Acknowledgements
I thank Xiaochen Bai and Rouslan Efremov for providing the micrographsand class averages of g-secretase and ryanodine receptor shown inFigure 1. I thank Richard Henderson, Nigel Unwin and Greg McMullan forcritical reading and comments on the manuscript. This work was supportedby the Medical Research Council Grant Number U105184322.
References and recommended readingPapers of particular interest, published within the period of review,have been highlighted as:
� of special interest�� of outstanding interest
1. Singer SJ, Nicolson GL: The fluid mosaic model of the structureof cell membranes. Science 1972, 175:720-731.
2. Blaurock AE, Stoeckenius W: Structure of the purple membrane.Nat New Biol 1971, 233:152-155.
3. Blaurock AE, Wilkins MH: Structure of retinal photoreceptormembranes. Nature 1972, 236:313-314.
www.sciencedirect.com
4. Henderson R, Unwin PN: Three-dimensional model of purplemembrane obtained by electron microscopy. Nature 1975,257:28-32.
5. Unwin PN, Henderson R: Molecular structure determination byelectron microscopy of unstained crystalline specimens. J MolBiol 1975, 94:425-440.
6.��
Dubochet J, Adrian M, Chang JJ, Homo JC, Lepault J,McDowall AW, Schultz P: Cryo-electron microscopy of vitrifiedspecimens. Q Rev Biophys 1988, 21:129-228.
This review describes the science behind freezing of thin films, a criticalstep in cryoEM and has beautiful illustrations.
7. Taylor KA, Glaeser RM: Electron diffraction of frozen, hydratedprotein crystals. Science 1974, 186:1036-1037.
8. Henderson R, Baldwin JM, Ceska TA, Zemlin F, Beckmann E,Downing KH: Model for the structure of bacteriorhodopsinbased on high-resolution electron cryo-microscopy. J Mol Biol1990, 213:899-929.
9. Grigorieff N, Ceska TA, Downing KH, Baldwin JM, Henderson R:Electron-crystallographic refinement of the structure ofbacteriorhodopsin. J Mol Biol 1996, 259:393-421.
10. Miyazawa A, Fujiyoshi Y, Unwin N: Structure and gatingmechanism of the acetylcholine receptor pore. Nature 2003,423:949-955.
11. Subramaniam S, Henderson R: Molecular mechanism ofvectorial proton translocation by bacteriorhodopsin. Nature2000, 406:653-657.
12. Unwin N: Acetylcholine receptor channel imaged in the openstate. Nature 1995, 373:37-43.
13. Raunser S, Walz T: Electron crystallography as a technique tostudy the structure on membrane proteins in a lipidicenvironment. Annu Rev Biophys 2009, 38:89-105.
14. Vinothkumar KR, Henderson R: Structures of membraneproteins. Q Rev Biophys 2010, 43:65-158.
15. Nietlispach D, Gautier A: Solution and NMR studies of polytopica-helical membrane proteins. Curr Opin Struct Biol 2011,21:497-508.
16.��
Henderson R: The potential and limitations of neutrons,electrons and X-rays for atomic resolution microscopy ofunstained biological molecules. Q Rev Biophys 1995, 28:171-193.
A forward thinking insightful review predicting the power and potentials ofcryoEM. This review and Ref. [6��] is highly recommended to anyonestarting on electron cryomicroscopy.
17.�
Rosenthal PB, Henderson R: Optimal determination of particleorientation, absolute hand, and contrast loss in single-particle electron cryomicroscopy. J Mol Biol 2003,333:721-745.
This article described the tilt-pair validation method and provides adetailed theoretical background on electron microscopy in general.
18. Grigorieff N, Harrison SC: Near-atomic resolutionreconstructions of icosahedral viruses from electron cryo-microscopy. Curr Opin Struct Biol 2011, 21:265-273.
19. Faruqi AR, McMullan G: Electronic detectors for electronmicroscopy. Q Rev Biophys 2011, 44:357-390.
20. Brilot AF, Chen JZ, Cheng A, Pan J, Harrison SC, Potter CS,Carragher B, Henderson R, Grigorieff N: Beam-induced motionof vitrified specimen on holey carbon film. J Struct Biol 2012,177:630-637.
21. Bai XC, Fernandez IS, McMullan G, Scheres SH: Ribosomestructures to near-atomic resolution from thirty thousandcryo-EM particles. Elife 2013, 2:e10046.
22. Campbell MG, Cheng A, Brilot AF, Moeller A, Lyumkis D, Veesler D,Pan J, Harrison SC, Potter CS, Carragher B et al.: Movies of ice-embedded particles enhance resolution in electron cryo-microscopy. Structure 2012, 20:1823-1828.
23. Li X, Mooney P, Zheng S, Booth CR, Braunfeld MB, Gubbens S,Agard DA, Cheng Y: Electron counting and beam-induced
Current Opinion in Structural Biology 2015, 33:103–114
112 Membranes
motion correction enable near-atomic-resolution single-particle cryo-EM. Nat Methods 2013, 10:584-590.
24. Vinothkumar KR, McMullan G, Henderson R: Molecularmechanism of antibody-mediated activation of beta-galactosidase. Structure 2014, 22:621-627.
25. van Heel M, Gowen B, Matadeen R, Orlova EV, Finn R, Pape T,Cohen D, Stark H, Schmidt R, Schatz M et al.: Single-particleelectron cryo-microscopy: towards atomic resolution. Q RevBiophys 2000, 33:307-369.
26. Frank J: Single-particle reconstruction of biologicalmacromolecules in electron microscopy – 30 years. Q RevBiophys 2009, 42:139-158.
27. Crowther RA: From envelopes to atoms: the remarkableprogress of biological electron microscopy. Adv Protein ChemStruct Biol 2010, 81:1-32.
28. Bai XC, McMullan G, Scheres SH: How cryo-EM isrevolutionizing structural biology. Trends Biochem Sci 2015,40:49-57.
29. Harris JR: Transmission electron microscopy in molecularstructural biology: a historical survey. Arch Biochem Biophys2014 http://dx.doi.org/10.1016/j.abb.11.2014.011.
30. Henderson R: Overview and future of single particle electroncryomicroscopy. Arch Biochem Biophys 2015 http://dx.doi.org/10.1016/j.abb.02.2015.036.
31.��
Cheng Y, Grigorieff N, Penczek PA, Walz T: A primer to single-particle cryo-electron microscopy. Cell 2015, 161:438-449.
An excellent recent review describing the steps underlying the techniqueof single particle cryoEM. The supplementary material has much moreinformation and should be read along with the main text.
32. Grigorieff N: Three-dimensional structure of bovineNADH:ubiquinone oxidoreductase (complex I) at 22 A in ice. JMol Biol 1998, 277:1033-1046.
33. Rhee KH, Scarborough GA, Henderson R: Domain movements ofplasma membrane H+-ATPase: 3D structures of two states byelectron cryo-microscopy. EMBO J 2002, 21:3582-3589.
34. Rubinstein JL, Walker JE, Henderson R: Structure of themitochondrial ATP synthase by electron cryomicroscopy.EMBO J 2003, 22:6182-6192.
35. Dickson VK, Silvester JA, Fearnley IM, Leslie AG, Walker JE: Onthe structure of the stator of the mitochondrial ATP synthase.EMBO J 2006, 25:2911-2918.
36. Henderson R: Avoiding the pitfalls of single particle cryo-electron microscopy: Einstein from noise. Proc Natl Acad Sci US A 2013, 110:18037-18041.
37.�
Yan Z, Bai XC, Yan C, Wu J, Li Z, Xie T, Peng W, Yin CC, Li X,Scheres SH et al.: Structure of the rabbit ryanodine receptorRyR1 at near-atomic resolution. Nature 2015, 517:50-55.
Refs. [37�,38�,39�] describe the structure of ryanodine receptor isolatedfrom the same source but differ in detergent/substitutes used for pur-ification. They reveal largely similar structure. These structures alsohighlight that medium to high-resolution structures can be obtained witha number of detergents and their substitutes.
38.�
Efremov RG, Leitner A, Aebersold R, Raunser S: Architecture andconformational switch mechanism of the ryanodine receptor.Nature 2015, 517:39-43.
See note in Ref. [37�].
39.�
Zalk R, Clarke OB, des Georges A, Grassucci RA, Reiken S,Mancia F, Hendrickson WA, Frank J, Marks AR: Structure of amammalian ryanodine receptor. Nature 2015, 517:44-49.
See note in Ref. [37�].
40. Westfield GH, Rasmussen SG, Su M, Dutta S, DeVree BT,Chung KY, Calinski D, Velez-Ruiz G, Oleskie AN, Pardon E et al.:Structural flexibility of the Gas a-helical domain in the b2-adrenoceptor Gs complex. Proc Natl Acad Sci U S A 2011,108:16086-16091.
41. Rasmussen SG, DeVree BT, Zou Y, Kruse AC, Chung KY,Kobilka TS, Thian FS, Chae PS, Pardon E, Calinski D et al.: Crystalstructure of the b2 adrenergic receptor-Gs protein complex.Nature 2011, 477:549-555.
Current Opinion in Structural Biology 2015, 33:103–114
42. Durr KL, Chen L, Stein RA, De Zorzi R, Folea IM, Walz T,McHaourab HS, Gouaux E: Structure and dynamics of AMPAreceptor GluA2 in resting, pre-open, and desensitized states.Cell 2014, 158:778-792.
43. Moiseenkova-Bell VY, Wensel TG: Hot on the trail of TRPchannel structure. J Gen Physiol 2009, 133:239-244.
44. Taylor CW, da Fonseca PC, Morris EP: IP3 receptors: the searchfor structure. Trends Biochem Sci 2004, 29:210-219.
45. Serysheva II, Ludtke SJ, Baker ML, Cong Y, Topf M, Eramian D,Sali A, Hamilton SL, Chiu W: Subnanometer-resolution electroncryomicroscopy-based domain models for the cytoplasmicregion of skeletal muscle RyR channel. Proc Natl Acad Sci U S A2008, 105:9610-9615.
46. Samso M, Wagenknecht T, Allen PD: Internal structure andvisualization of transmembrane domains of the RyR1 calciumrelease channel by cryo-EM. Nat Struct Mol Biol 2005, 12:539-544.
47.�
Lau WC, Rubinstein JL: Subnanometre-resolution structure ofthe intact Thermus thermophilus H+-driven ATP synthase.Nature 2011, 481:214-218.
Before the introduction of direct electron detectors, this structure wasdetermined to <10 A and was an important contribution to field ofcryoEM.
48. Becker T, Bhushan S, Jarasch A, Armache JP, Funes S, Jossinet F,Gumbart J, Mielke T, Berninghausen O, Schulten K et al.:Structure of monomeric yeast and mammalian Sec61complexes interacting with the translating ribosome. Science2009, 326:1369-1373.
49. Frauenfeld J, Gumbart J, Sluis EO, Funes S, Gartmann M,Beatrix B, Mielke T, Berninghausen O, Becker T, Schulten K et al.:Cryo-EM structure of the ribosome-SecYE complex in themembrane environment. Nat Struct Mol Biol 2011, 18:614-621.
50. Gogala M, Becker T, Beatrix B, Armache JP, Barrio-Garcia C,Berninghausen O, Beckmann R: Structures of the Sec61complex engaged in nascent peptide translocation ormembrane insertion. Nature 2014, 506:107-110.
51. Menetret JF, Hegde RS, Aguiar M, Gygi SP, Park E, Rapoport TA,Akey CW: Single copies of Sec61 and TRAP associate with anontranslating mammalian ribosome. Structure 2008, 16:1126-1137.
52. Pfeffer S, Dudek J, Gogala M, Schorr S, Linxweiler J, Lang S,Becker T, Beckmann R, Zimmermann R, Forster F: Structure ofthe mammalian oligosaccharyl-transferase complex in thenative ER protein translocon. Nat Commun 2014, 5:3072.
53. Park E, Menetret JF, Gumbart JC, Ludtke SJ, Li W, Whynot A,Rapoport TA, Akey CW: Structure of the SecY channel duringinitiation of protein translocation. Nature 2014, 506:102-106.
54. Wickles S, Singharoy A, Andreani J, Seemayer S, Bischoff L,Berninghausen O, Soeding J, Schulten K, van der Sluis EO,Beckmann R: A structural model of the active ribosome-boundmembrane protein insertase YidC. Elife 2014, 3:e50303.
55.��
Liao M, Cao E, Julius D, Cheng Y: Structure of the TRPV1 ionchannel determined by electron cryo-microscopy. Nature2013, 504:107-112.
A biologically important structure determined to high-resolution withoutthe need for crystals. A landmark in single particle cryoEM of membraneproteins.
56.�
Lu P, Bai XC, Ma D, Xie T, Yan C, Sun L, Yang G, Zhao Y, Zhou R,Scheres SH et al.: Three-dimensional structure of humangamma-secretase. Nature 2014, 512:166-170.
Refs. [56�,57�,58�,59�,60�,61�,62�,63�,64�] are recent membrane proteinstructures determined by single particle cryoEM. They vary in size andcomplexity and provide a glimpse on what can be expected from singleparticle cryoEM of membrane proteins in the near future.
57.�
Meyerson JR, Kumar J, Chittori S, Rao P, Pierson J, Bartesaghi A,Mayer ML, Subramaniam S: Structural mechanism of glutamatereceptor activation and desensitization. Nature 2014, 514:328-334.
See note in Ref. [56�].
www.sciencedirect.com
Membrane protein structures without crystals Vinothkumar 113
58.�
Vinothkumar KR, Zhu J, Hirst J: Architecture of mammalianrespiratory complex I. Nature 2014, 515:80-84.
See note in Ref. [56�].
59.�
Kim J, Wu S, Tomasiak TM, Mergel C, Winter MB, Stiller SB,Robles-Colmanares Y, Stroud RM, Tampe R, Craik CS et al.:Subnanometre-resolution electron cryomicroscopy structureof a heterodimeric ABC exporter. Nature 2015, 517:396-400.
See note in Ref. [56�].
60.�
Allegretti M, Klusch N, Mills DJ, Vonck J, Kuhlbrandt W,Davies KM: Horizontal membrane-intrinsic alpha-helices in thestator a-subunit of an F-type ATP synthase. Nature 2015 http://dx.doi.org/10.1038/nature.1418.5.
See note in Ref. [56�].
61.�
Rawson S, Phillips C, Huss M, Tiburcy F, Wieczorek H, Trinick J,Harrison MA, Muench SP: Structure of the Vacuolar H+-ATPaserotary motor reveals new mechanistic insights. Structure 2015,23:461-471.
See note in Ref. [56�].
62.�
Jiang J, Pentelute BL, Collier RJ, Zhou ZH: Atomic structure ofanthrax protective antigen pore elucidates toxintranslocation. Nature 2015 http://dx.doi.org/10.1038/nature.142.7.
See note in Ref. [56�].
63.�
Paulsen CE, Armache JP, Gao Y, Cheng Y, Julius D: Structure ofthe TRPA1 ion channel suggests regulatory mechanisms.Nature 2015, 520:511-517.
See note in Ref. [56�].
64.�
Zhao J, Benlekbir S, Rubinstein JL: Electron cryomicroscopyobservation of rotational states in a eukaryotic V-ATPase.Nature 2015, 521:241-245.
See note in Ref. [56�].
65. Sun L, Zhao L, Yang G, Yan C, Zhou R, Zhou X, Xie T, Zhao Y,Wu S, Li X et al.: Structural basis of human gamma-secretase assembly. Proc Natl Acad Sci U S A 2015,112:6003-6008.
66. Voorhees RM, Fernandez IS, Scheres SH, Hegde RS: Structure ofthe mammalian ribosome-Sec61 complex to 3.4 A resolution.Cell 2014, 157:1632-1640.
67. Lyumkis D, Brilot AF, Theobald DL, Grigorieff N: Likelihood-based classification of cryo-EM images using FREALIGN. JStruct Biol 2013, 183:377-388.
68. Scheres SH, Gao H, Valle M, Herman GT, Eggermont PP, Frank J,Carazo JM: Disentangling conformational states ofmacromolecules in 3D-EM through likelihood optimization.Nat Methods 2007, 4:27-29.
69. Liao HY, Frank J: Definition and estimation of resolutionin single-particle reconstructions. Structure 2010,18:768-775.
70. Henderson R, Chen S, Chen JZ, Grigorieff N, Passmore LA,Ciccarelli L, Rubinstein JL, Crowther RA, Stewart PL,Rosenthal PB: Tilt-pair analysis of images from a range ofdifferent specimens in single-particle electroncryomicroscopy. J Mol Biol 2011, 413:1028-1030.
71. Chen S, McMullan G, Faruqi AR, Murshudov GN, Short JM,Scheres SH, Henderson R: High-resolution noise substitution tomeasure overfitting and validate resolution in 3D structuredetermination by single particle electron cryomicroscopy.Ultramicroscopy 2013, 135:24-35.
72. Scheres SH: RELION: implementation of a Bayesian approachto cryo-EM structure determination. J Struct Biol 2012,180:519-530.
73. Scheres SH, Chen S: Prevention of overfitting in cryo-EMstructure determination. Nat Methods 2012, 9:853-854.
74. Kucukelbir A, Sigworth FJ, Tagare HD: Quantifying the localresolution of cryo-EM density maps. Nat Methods 2014,11:63-65.
75. Brown A, Long F, Nicholls RA, Toots J, Emsley P, Murshudov G:Tools for macromolecular model building and refinement into
www.sciencedirect.com
electron cryo-microscopy reconstructions. Acta Crystallogr D:Biol Crystallogr 2015, 71:136-153.
76. Wang RY, Kudryashev M, Li X, Egelman EH, Basler M, Cheng Y,Baker D, DiMaio F: De novo protein structure determinationfrom near-atomic-resolution cryo-EM maps. Nat Methods 2015http://dx.doi.org/10.1038/nmeth.3287.
77. DiMaio F, Song Y, Li X, Brunner MJ, Xu C, Conticello V, Egelman E,Marlovits TC, Cheng Y, Baker D: Atomic-accuracy models from4.5-A cryo-electron microscopy data with density-guidediterative local refinement. Nat Methods 2015 http://dx.doi.org/10.1038/nmeth.3286.
78. Wang Z, Hryc CF, Bammes B, Afonine PV, Jakana J, Chen DH,Liu X, Baker ML, Kao C, Ludtke SJ et al.: An atomic model ofbrome mosaic virus using direct electron detection and real-space optimization. Nat Commun 2014, 5:4808.
79. Wu S, Avila-Sakar A, Kim J, Booth DS, Greenberg CH, Rossi A,Liao M, Li X, Alian A, Griner SL et al.: Fabs enable single particlecryoEM studies of small proteins. Structure 2012, 20:582-592.
80. Kuhlbrandt W: Biochemistry. The resolution revolution. Science2014, 343:1443-1444.
81. Fischer N, Neumann P, Konevega AL, Bock LV, Ficner R,Rodnina MV, Stark H: Structure of the E. coli ribosome-EF-Tucomplex at <3 A resolution by Cs-corrected cryo-EM. Nature2015 http://dx.doi.org/10.1038/nature.1427.5.
82. Grant T, Grigorieff N: Measuring the optimal exposure for singleparticle cryo-EM using a 2.6 A reconstruction of rotavirus VP6.Elife 2015:4.
83. Bartesaghi A, Merk A, Banerjee S, Matthies D, Wu X, Milne JL,Subramaniam S: Electron microscopy 2.2 A resolution cryo-EMstructure of b-galactosidase in complex with a cell-permeantinhibitor. Science 2015, 348:1147-1151.
84. Scheres SH: Beam-induced motion correction for sub-megadalton cryo-EM particles. Elife 2014, 3:e50366.
85. Leis A, Rockel B, Andrees L, Baumeister W: Visualizing cells atthe nanoscale. Trends Biochem Sci 2009, 34:60-70.
86. Schur FK, Hagen WJ, Rumlova M, Ruml T, Muller B,Krausslich HG, Briggs JA: Structure of the immature HIV-1capsid in intact virus particles at 8.8 A resolution. Nature 2015,517:505-508.
87. Briggs JA: Structural biology in situ — the potential ofsubtomogram averaging. Curr Opin Struct Biol 2013,23:261-267.
88. Davies KM, Anselmi C, Wittig I, Faraldo-Gomez JD, Kuhlbrandt W:Structure of the yeast F1Fo-ATP synthase dimer and its role inshaping the mitochondrial cristae. Proc Natl Acad Sci U S A2012, 109:13602-13607.
89. Zuber B, Unwin N: Structure and superorganization ofacetylcholine receptor-rapsyn complexes. Proc Natl Acad Sci US A 2013, 110:10622-10627.
90. Pfeffer S, Woellhaf MW, Herrmann JM, Forster F: Organization ofthe mitochondrial translation machinery studied in situ bycryoelectron tomography. Nat Commun 2015, 6:6019.
91.�
Russo CJ, Passmore LA: Electron microscopy: ultrastable goldsubstrates for electron cryomicroscopy. Science 2014,346:1377-1380.
Refs. [91�,92�] describe alternative support material for cryoEM grids. InRef. [91�], gold substrate is shown to reduce some beam-induced move-ment. In Ref. [92�], normal holey carbon grids coated with gold ischemically modified so that better particle distribution can be obtained.
92.�
Meyerson JR, Rao P, Kumar J, Chittori S, Banerjee S, Pierson J,Mayer ML, Subramaniam S: Self-assembled monolayersimprove protein distribution on holey carbon cryo-EMsupports. Sci Rep 2014, 4:7084.
See note in Ref. [91�].
93. Dai W, Fu C, Khant HA, Ludtke SJ, Schmid MF, Chiu W: Zernikephase-contrast electron cryotomography applied to marinecyanobacteria infected with cyanophages. Nat Protoc 2014,9:2630-2642.
Current Opinion in Structural Biology 2015, 33:103–114
114 Membranes
94. Danev R, Buijsse B, Khoshouei M, Plitzko JM, Baumeister W: Voltapotential phase plate for in-focus phase contrast transmissionelectron microscopy. Proc Natl Acad Sci U S A 2014, 111:15635-15640.
95.�
Cao E, Liao M, Cheng Y, Julius D: TRPV1 structures in distinctconformations reveal activation mechanisms. Nature 2013,504:113-118.
Current Opinion in Structural Biology 2015, 33:103–114
A companion article to Ref. [55��] describing structures of TRPV1 channelin different states highlighting the power of cryoEM.
96. Baradaran R, Berrisford JM, Minhas GS, Sazanov LA: Crystalstructure of the entire respiratory complex I. Nature 2013,494:443-448.
www.sciencedirect.com