Available via license: CC BY-NC-SA 4.0
Content may be subject to copyright.
Architecture of the
Xenopus
Nuclear Pore Complex Revealed by
Three-Dimensional Cryo-Electron Microscopy
Christopher
W. Akey and Michael Radermacher*
Department of Biophysics, Boston University School of Medicine, Boston, Massachusetts 02118-2394; and * Wadsworth Center
for Laboratories and Research, New York State Department of Health, Albany, New York 12201-509
Abstract. The nuclear pore complex spans the nuclear
envelope and functions as a macromolecular trans-
porter in the ATP-dependent process of nucleo-
cytoplasmic transport. In this report, we present three
dimensional (3D) structures for both membrane-
associated and detergent-extracted Xenopus NPCs, im-
aged in frozen buffers by cryo-electron microscopy. A
comparison of the differing configurations present in
the 3D maps suggests that the spokes may possess an
intrinsic conformational flexibility. When combined
with recent data from a 3D map of negatively stained
NPCs (Hinshaw, J. E., B. O. Carragher, and R. A.
Milligan. 1992. Cell. 69:1133-1141), these observa-
tions suggest a minimal domain model for the
spoke-ring complex which may account for the ob-
served plasticity of this assembly. Moreover, lumenal
domains in adjacent spokes are interconnected by
radial arm dimers, forming a lumenal ring that may be
responsible for anchoring the NPC within the nuclear
envelope pore. Importantly, the NPC transporter is
visualized as a centrally tapered cylinder that spans
the entire width of the NPC, in a direction normal to
the nuclear envelope. The central positioning, tripar-
tite structure, and hollow nature of the transporter
suggests that it may form a macromolecular transport
channel, with a globular gating domain at each end.
Finally, the packing of the transporter within the
spokes creates a set of eight internal channels that may
be responsible, in part, for the diffusion of ions and
small molecules across the nuclear envelope.
T hE nuclear pore complex (NPC) ~ resides within a
pore formed by the fusion of the inner and outer nu-
clear membranes and is tethered to the nuclear lam-
ina. Therefore, the NPC is ideally situated to mediate the bi-
directional exchange of small molecules and catalyze the
active transport of proteins, cellular and viral RNPs (15, 25,
29, 48). Accumulating evidence suggests that nucleocyto-
plasmic transport is a singular reaction in that: (a) transport
is bidirectional yet vectorial for a particular substrate at a
given instant; (b) transport occurs through the center of the
NPC as visualized by thin sections of Balbiani ring mRNPs
transiting the NPC (61); (c) an unusually broad range of sub-
strate types and sizes are transported; (d) substrates need not
be deformed in passage (21); (e) the import of many
karyophilic proteins is dependent on a positively charged nu-
clear localization signal (NLS), that is bipartite in nucleo-
plasmin and SWI5 (43, 56); and (f) import of karyophilic
proteins requires at least two cytosolic factors (1, 44, 45) and
involves specific binding to O-linked N-acetyl glucosamine
containing nucleoporins (58).
1. Abbreviations used in this paper:
CP, cytoplasmic particles; CR/NR, cy-
toplasmic ring/nucleoplasmic ring; CTF, contrast transfer function; ISR,
inner spoke ring; NLS, nuclear localization signal; NPC, nuclear pore com-
plex; RA, radial arm; 3D, three dimensional.
Substrates transported by the NPC are chemically diverse
and include: proteins, mRNPs, snRNPs, pre-ribosomal
subunits, and tRNAs. Recent data suggests that they use a
common transport machine located within the NPC, as both
import and export reactions for these substrates demonstrate
ATP-dependent saturation kinetics (9, 14, 30, 41, 46, 65) and
manifest similar inhibition profiles (13, 14, 19, 24, 41). In-
terestingly, protein import and the export of nucleic acids ap-
pears to occur through common NPCs (17). In addition, the
transport of proteins and mRNPs are topologically similar,
as both processes use a central channel within the NPC that
constrains substrates to a maximum diameter of 200-250/~
during translocation (28, 61, 23, 40). The identity and struc-
ture of the macromolecular transport machine located cen-
trally within the NPC has not been established.
The NPC is a cylindrical assembly with octagonal sym-
metry and a mass of '~125 Mda (18, 52). Disassembly
studies (52, 62) and quantitative image analyses of en face
and side-on projections have suggested that the NPC is com-
prised of a set of thin rings which sandwich a more massive
spoke assembly (3, 52, 62). Projections of frozen-hydrated
specimen have revealed pairs of radial arms that inter-
connect adjacent spokes and a central ringlike transporter
(3), that may be capable of forming a gated transport channel
for substrates (4, 8). The precise symmetry of the NPC
proper may vary depending on structural changes induced
© The Rockefeller University Press, 0021-9525/93/07/1/19 $2.00
The Journal of Cell Biology, Volume 122, Number 1, July 1993 1-19 1
during enucleation and sample preparation (3). The NPC
may he either eightfold symmetric about an axis perpendicu-
lar to the nuclear envelope (3, 18, 28, 52) or "o822 symmetric
with an approximate twofold axis relating the oppositely fac-
ing halves of the spoke-ring complex (3, 4, 42, 62). Re-
cently, Hinshaw and co-workers (34) have completed a single
particle three-dimensional (3D) reconstruction from deter-
gent-extracted (dform) and detached
Xenopus
NPCs with
'o822 symmetry. As visualized in negative stain, the mini-
mal building block of the spoke-ring complex is composed
of five domains including two domains within each of the
averaged thin rings and three domains in each half-spoke. Pe-
ripheral assemblies, such as cytoplasmic particles, filaments
(28, 53, 54, 55, 59, 62), and nucleoplasmic cages/baskets
(28, 36, 54, 55) were not visualized in these detached spec-
imens.
In this report, we present data from independent 3D
reconstructions of
Xenopus
NPCs tethered to the nuclear
lamina in the presence or absence of the nuclear envelope.
The maps provide insights on issues concerning: (a) the min-
imal domain structure of the spoke-ring complex and its ob-
served conformational flexibility; (b) the possible role of the
NPC transporter in nucleocytoplasmic transport; (c) the po-
sitions of internal channels that may function as diffusion
pores; and (d) interactions between the NPC and the encir-
cling nuclear envelope.
Materials and Methods
Specimen Preparation and Image Processing
The maintenance of amphibians, isolation of oocytes, and preparation of
frozen-hydrated nuclear envelope spreads for cryo-electron microscopy
were carried out as described previously (3). Two microscopes were used
in the data collection, a Philips EM400T at the Deptartment of Cell Biology,
Stanford University Medical School (Stanford, CA) and an EM400 in the
Structural Studies Division at the Medical Research Council Laboratory of
Molecular Biology. Both microscopes were modified with objective lens
current meters by Dr. C. Toyoshima (Department of Biological Sciences,
Tokyo Institute of Technology, Tokyo, Japan) which were used to adjust the
zero and tilted specimen pair z-heights, such that a consistent objective lens
current was obtained, thereby ensuring a uniform magnification. Micro-
graphs were scanned on a flatbed densitometer manufactured by Joyce,
Loebl and Co. Ltd. (Gateshead, England), that was extensively modified
at the LMB. The effective step sizes on the images were 24.6 and 26.7/~
depending on the initial magnification. 3D image processing used SECRET,
a 3D reconstruction method for single particles based on a conical tilt geom-
etry (50, 51), implemented within the SPIDER software package (26). A
suite of roughly 100 SPIDER procedures and sub-procedures were written
and integrated into a single automated 2D/3D program set, ALIGN___EM,
based in part on a previous set of programs written by M. Radermacher
(New York State Department of Health, Albany, NY) (RCDOALL).
ALIGN_..EM runs five deep within a batch job and provides a highly struc-
tured environment for data and document files during the analysis of large
datasets. The current version has over 40 steps in a sequential menu driven
format that covers 2D/3D alignments and the preparation of computed 3D
maps for subsequent analysis and presentation.
Specific modifications were made in the procedures used to align tilted
particles, to accommodate the low-contrast images of frozen-hydrated spec-
imens. In particular, larger areas of 256 × 256 were windowed out around
the zero and tilted particles and used only in the alignment of the un-
tilted/tilted image pairs, to boost the signal of the cross-correlation peak.
Images in 128 × 128 format were used routinely for all other calculations.
In addition, all tilted particles used in the 3D reconstructions had the first
minima of their contrast transfer functions (CTFs) outside the nominal reso-
lution range. Cylindrical averages of the tilted and aligned particles were
used to judge the accuracy of alignment along with the convergence of the
step sizes assigned to each particle during refinement. Further improve-
merits in the tilt centration parameters were achieved by iterative refinement
of each tilted image against a projection of an unweighted 3D reconstruc-
tion, calculated from the entire dataset (Radermacher, 1992). In this ap-
proach the tilted image under refinement (with negative contrast) was re-
moved from the trial 3D reconstruction by back projection, before a
centration alignment between the original tilted image (with positive con-
tras0 and the equivalent projection calculated from the 3D volume. Cor-
rected xy shifts were then used to reinsert the tilted image into the 3D vol-
ume by back projection and the process continued for all images in the
dataset through two complete cycles. This refinement resulted in small but
significant changes in the xy shifts obtained from cross-correlational align-
ments of the zero/tilted image pairs.
Datasets with tilt angles of 34 ° (mform) and 42 ° (dform) were used to
minimize problems from nearest neighbor overlap in the tilted specimens;
however this resulted in rather large missing cones. Therefore, a real
space/Fourier space method (POCs/solvent flattening; 12) was employed
that uses volume constraints to iteratively extend the molecular transform
into the cone of missing information. This approach was incorporated into
SPIDER by M. Radermacher and carded out as follows. A weighted 3D
reconstruction of the complete dataset was used to generate a 3D mask of
the NPC using a combination of thresholding and low pass filtering. The
mask was then applied to the 3D volume in real space and the 3D Fourier
transform calculated. The resulting molecular transform in the missing cone
was left untouched between cycles, but the calculated Fourier in the mea-
sured area of the molecular transform was replaced with the original molec-
ular transform (from the unmasked object) and the resulting hybrid back
Fourier transformed. This procedure was carded out for two to three cycles
before defining a smaller mask and repeating the process. The final molecu-
lar volume of the mask was roughly 1.5 times the calculated molecular vol-
ume of the NPC based on STEM mass measurements (52). This cutoff was
used to compensate for the unknown mass and undetermined occupancy of
the cytoplasmic particles and material "trapped" within the transporter (4).
Solvent flattening produced 3D maps whose peak distribution and relative
intensities were unaffected in the xy plane sections; however, the modified
maps demonstrated a higher contrast relative to background and a net
shrinkage along the z-axis of 12-16%, resulting in height/diameter ratios
that agree with side-on views (3). The series of 3D masks were used to
refine half-datasets in parallel and the resulting 3D maps used to calculate
phase residuals (27). Phases within the missing cone were random at the
beginning of POCs refinement, but converged to give residuals similar to
those in the original measured regions of the molecular transform (,,o100]~).
Final 3D volumes were merged from the half-datasets and low pass filtered
at 80/~ resolution to compensate for doubling the amount of data. The POCs
method worked well because the original NPCs are not appreciably flat-
tened in the frozen-hydrated state. Moreover, an excellent fit of common
features was obtained when the central sections in each map were scaled
together, but regions within the scaled maps corresponding to the thin rings
in the mform were 6-10% smaller in diameter than their counterparts from
the dform. Multicolor surface rendering techniques were used to highlight
specific regions within the 3D maps using an algorithm written by M.
Radermacber to create a composite surface for display on WEB, a Motif
display program written by A. Leith (New York State Department of
Health, Albany, NY). Cutoffs were chosen to illustrate reproducible fea-
tures and generally fell at the boundaries of steep density gradients. Un-
known parameters such as the absolute occupancy of cytoplasmic particles
and material caught in transit within the transporter, coupled with uncer-
tainties arising from the loss of the lowest resolution frequencies of the CTF,
prevented us from placing the surface maps on an absolute volume scale
commensurate with the measured mass (52).
Results
We have used cryo-electron microscopy of frozen-hydrated
specimens to visualize the architecture of the NPC in 3D.
This approach minimizes perturbations that can occur dur-
ing negative staining and drying of large specimens. These
methods have produced both pseudosymmetric ('o822) and
asymmetric (C8) averages of the NPC associated with either
the nuclear envelope or the nuclear lamina (3, 4, 7). We have
used the Single Exposure random Conical tilt Reconstruc-
tion Technique (SECRET; 50, 51) which uses a pair of micro-
graphs, one tilted and one untilted, to obtain the orientation
angles necessary to calculate 3D maps from tilted projec-
tions. Maps of membrane-associated (mform) and detergent-
The Journal of Cell Biology, Volume 122, 1993 2
Figure 1. Representative areas of electron micrographs of zero/tilt pairs of frozen-hydrated Xenopus NPCs are shown. Note that the circular
NPCs become elliptical in the tilted images and show evidence of the triple ring morphology. (a and b) Zero and 34 ° tilt images of
membrane-associated NPCs. Dark areas are protein with the exception of crystalline ice particles (c). Large particles associated with the
nuclear envelope are visible in the spaces between NPCs (black arrowheads in a). The position of the tilt axis is indicated by the white
arrow in b. NPCs with central transporters are indicated with white arrowheads. (c and d) Zero and 42 ° tilt image pair recorded from
a Xenopus macronucleus extracted with TX-lfI0. The tilt axis position is marked by the white arrow in d. Ringlike NPC transporters are
present with high_occupancy in the zero tilt NPCs (white arrowheads in c) and the lamina is present as a fine meshwork between the NPCs
(L). Bar, 2,000 A.
Akey and Radermacher Architecture of the Xenopus Nuclear Pore Complex 3
Figure 2. Projection maps of both untilted (eightfold averaged) and tilted (cylindrically averaged) frozen-hydrated NPCs; contrast is reversed
with protein and strongly scattering material white. The maximum diameter of the untilted NPCs excluding the radial arms is 1,200/~.
(a) Map from 196 Xenopus membrane-associated NPCs; the spokes (S), radial arms (RA), and the membrane border (M) are indicated.
(b) Map of Xenopus detergent-extracted form calculated from 284 NPCs. Substructure is labeled as above with the exception of the central
NPC transporter (T), which appears in this view as an averaged plug. The positions of approximate twofold axes aligned with the spokes
and radial arms are indicated by white arrows. (c) A map of 268 NPCs obtained from a second detergent-extracted nuclear ghost is shown.
Note the similarity to the map in b. (d) A cylindrical average obtained from tilted and aligned NPCs (n = 196) associated with the nuclear
envelope is shown. The ring of spokes and the membrane border are present as an elliptical doublet; the thin cytoplasmic and nucleoplasmic
rings (CR and NR) are labeled. (e) A cylindrical average of detergent-extracted NPCs (n = 284) is shown. The NPC transporter forms
a broad band of density along the horizontal axis. (f) Map from 284 Necturus NPCs associated with a detergent-extracted nucleus isolated
from an animal with unhealthy oocytes. Note the weak inner spoke ring (ISR) formed by the slewed inner spoke domains. Density within
the spokes is somewhat asymmetric. Bar, 500 ./~.
extracted (dform) NPCs from Xenopus were obtained by this
method and are described in the following sections. At this
stage, the resolution is inadequate to precisely identify
subunit boundaries within the molecular envelopes of the
maps. However, a detailed side-by-side comparison allows a
qualitative description of the minimal number of domains
which comprise the framework of the spoke-ring complex.
This analysis confirms the 3D map of Hinshaw and co-
workers (34) and extends our understanding of the confor-
mationally flexible spoke-ring assembly and its interactions
with the central NPC transporter.
NPCs Associated with the Nuclear Envelope
Representative areas of a micrograph pair recorded at 0 ° and
34 ° from a spread Xenopus nuclear envelope are shown in
Fig. 1, a and b. The mform NPCs used for the 3D recon-
struction were circular with a distinct eightfold symmetry
and demonstrated a ring of density centered at a radius of
"~420-430/~ that traverses the centers of individual spokes
in projection (3, 4, 36, 62). The specimens are unstained;
hence, protein dominates the scattering and is dark. The tilt
axis (34 °) is located parallel to the white arrow in Fig. 1 b,
along the major axis of the elliptical NPCs in the tilted im-
age. Although a few NPCs retain their ringlike central trans-
porters (white arrowheads Fig. 1 a), the majority have lost
them during specimen preparation. In addition, large par-
ticles located between the NPCs (indicated by black ar-
rowheads) may correspond to either detached cytoplasmic
particles observed in the 3D maps or possibly to membrane-
associated ribosomes (62).
Averaged projection maps resulting from computer-based
alignments of untilted and tilted (34 °) NPCs are shown in
Fig. 2, a and d. Contrast has been reversed relative to the
micrographs; protein is now white. The averaged map calcu-
lated from the untilted NPCs is similar to maps obtained pre-
viously from
Xenopus (3). For example, the eight spokes and
The Journal of Cell Biology, Volume 122, 1993 4
Figure 3.
Diagram of a model cylinder with the x-, y- and z-axes
indicated. The slices described in the text are shown including an
xy slice cut perpendicular to the z-axis (the dark plane) and a z-slice
which corresponds to a cross-section cut parallel to and containing
the z-axis. As an example, a z slice is shown cut along an arbitrary
radial axis denoted by the angle theta. The z slices described in the
text were chosen as follows. The spoke z slice was chosen to bisect
individual spokes along their approximate twofold axis. Radial ann
z slices were obtained by rotating the plane of the cross-section by
22.5 ° clockwise, thereby bisecting the radial arms and the inner
spoke ring.
their attachments to the radial arms are visible. In addition'
the spokes are bisected by a ring of density that may arise
from the border of the fused inner and outer nuclear mem-
branes. The transporter is absent from the 2D map; instead
a central cavity is present with a diameter of ,x,440/~ (Fig.
2 a and see Fig. 4). A cylindrically averaged map was calcu-
lated from the tilted and aligned NPCs (Fig. 2 d). A cylindri-
cal average is obtained because all possible rotational orien-
tations, relative to the central eightfold axis, are present in
the corresponding untilted NPCs. Hence, after tilting to 34 °
and averaging over 196 NPCs a rotationally "blurred" image
is obtained without subunit details as viewed from above.
Importantly, the cylindrical average provides direct evidence
that the cytoplasmic and nucleoplasmic rings are present
within the specimens. Moreover, the membrane border and
spoke subunits appear to be split into concentric elliptical
rings that "encircle" the central cavity. After the tilted NPCs
were aligned to a common origin, a 3D map was calculated
by back projection. The cytoplasmic and nucleoplasmic
sides were then identified as follows. First, features within
the map could be correlated directly with their counterparts
in the 3D map of the dform. This form possesses a distinct
handedness in projection that has been used previously to as-
sign a unique viewing direction (3). Second, the cytoplasmic
ring in the maps was "denser" due to the presence of the cyto-
plasmic particles. This observation is in agreement with
previous observations (33, 52, 55, 62).
After refinement, differing views of domains within the 3D
maps were obtained by taking appropriate xy slices or z
slices (cross-sections) of the 3D volume. The orientation of
xy slices cut perpendicular to the z-axis is shown in Fig. 3,
within an idealized cylindrical volume. In Fig. 3, the z-axis
would correspond to the central eightfold symmetry axis of
the NPC. A set of xy-slices for the mform is shown in Fig.
4 (a-f),
and the spacings of these slices are shown in a side
view of a surface model to orient the reader (Fig. 4 g). The
3D map demonstrates an approximate twofold axis of sym-
metry within the central section (arrows 1 and 2 in Fig. 4 d)
that is maintained within the spokes. Symmetry axis 1 is cen-
tered on each spoke while symmetry axis 2 is rotated 22.5 °
clockwise from axis 1 and bisects the radial arms. In total,
both approximate twofold axes are repeated eight times
around the cylindrical NPC. The approximate twofold sym-
metry of the spokes is shown by comparing xy-slices in Fig.
4, c and f, which are related by a 180 ° rotation about the cen-
ter of the spokes. However, the thin rings are different at this
resolution (80-100A) with the cytoplasmic ring having eight
paired domains (labeled 1 and 2) and an inwards pointing as-
pect (see
asterisk
in Fig. 4 b), while the nucleoplasmic ring
has eight contiguous single domains and is less intense (com-
pare Fig. 4, b and e). Furthermore, the departure from per-
fect twofold symmetry is most marked at the cytoplasmic
ring which is decorated with the remnants of cytoplasmic
particles or collapsed filaments (see Fig. 4 a). The centers
of the cytoplasmic particles are offset counterclockwise by
,'o17 ° from the putative spoke twofold axis (symmetry axis
"1" in Fig. 4 d).
Inspection of the central xy slice in Fig. 4 d reveals three
density peaks within each spoke including: the inner domain
(IS), a central domain (CS), and the lumenal domain (LS).
Furthermore, certain spoke domains join together to form
larger concentric structures. For example, the inner spoke
domains are interconnected to form an inner spoke ring
(ISR) and the lumenal domains penetrate into the nuclear lu-
men where they are linked by radial arm dimers (RA), form-
ing a lumenal ring. Additional features are visible in z-slices,
as defined in Fig. 3, that correspond to planar cross-sections
cut parallel to the z-axis along the arrows labeled 1 and 2 in
Fig. 4 d. In Fig. 5 a, the z-slice bisects a spoke showing the
vertical and radial density variations within oppositely fac-
ing spokes. Interestingly, the inner spoke domains make
their strongest contacts to the rodlike density that contains
the central domain at the upper and lower extremities, rather
than at the midline. Other contributions to the vertical rod-
shaped density result from the juxtaposition of the outer ver-
tical domains (Vo) and the central domain with the nuclear
envelope border. The lumenal spoke domain protrudes from
the spoke surface into the lumen of the nuclear envelope. A
second cross-section cut 22.5 ° away and bisecting the radial
arms, shows portions of five domains including the ISR, the
cytoplasmic and nucleoplasmic rings (CR/NR), the cyto-
plasmic particles (CP), and the radial arms (RA). A weaker
band of density (indented in the middle) runs vertically along
the outer surface of the inner spoke ring and, at this resolu-
tion, connects the cytoplasmic and nncleoplasmic thin rings.
This density may be attributed in part to the border of the
nuclear envelope (M). In Fig. 4 d, the central domains (CS)
appear to traverse this ring of density. However, structural
rearrangements within the inner spoke ring may also contrib-
ute to the "membrane" density at this radius.
Color-coded surface views of the 3D map are presented in
Akey and Radermacher
Architecture of the Xenopus Nuclear Pore Complex 5
Figure
4. A montage is shown of averaged xy slices taken from the final 3D map of the membrane-associated NPC. The complete volume
has 64 slices spaced 25-A apart along the z-axis. A sideview of the 3D structure is shown in g as a surface model, with the slice positions
and external domains indicated. This map maintains approximate twofold symmetry over most of the spoke assembly, in particular the
central section (d) and slices in c andfshow the expected twofold symmetric relationship. (a) Ring of cytoplasmic particles (CP): aver-
age of slices 13-17. (b) Thin cytoplasmic ring
(CR)
with two circumferential domains and a smaller inwards pointing domain
(asterisk):
average of slices 19-22. (c) Cytoplasmic region of the spoke assembly above the lumenal spoke domain: average of slices 26-27. (d) Central
slice of the 3D map with the spoke and radial ann twofold axes shown
(white arrows 1 and 2).
Positions of the radial arms (RA), lumenal
spoke domains (LS), membrane (M), central spoke domain
(CS),
inner spoke domain
(IS),
and inner spoke ring
(ISR) are
labeled. This
section displays approximate twofold symmetry. (e) Nucleoplasmic thin ring with elongated subunits: average of slices 42-45. (f) Lower
region of the spoke assembly that is related by approximate twofold symmetry to the slice in c: average of slices 3%38. (g) Side view
of surface map showing the rudimentary cytoplasmic particles
(CP),
cytoplasmic and nucleoplasmic thin rings
(CR and NR),
lumenal
domains (LS), and the diagonally bridging radial arm dimers (RA).
The Journal of Cell Biology, Volume 122, 1993 6
Figure 5.
Cross-sections (z
slices) taken from 3D maps of
mforrn and dform NPCs after
refinement by POCs/solvent
flattening, a and c are slices
cut along the putative spoke
twofold axes; b and d are
slices cut along along the
putative radial arm twofold
axes. b and d are related to a
and c by a rotation of 22.5 °
around the central eightfold
axis. (a) A z slice which
bisects the spokes of the
membrane-associated form is
shown. Domains within the
spokes are labeled including:
inner spoke domain
(IS),
cen-
tral domain
(CS),
lumenal do-
main (LS), the inner vertical
domain (V/), and the outer
vertical domain (I6). An area
of
weak
interaction
between
the inner spoke domain and
the
central spoke
domain
is la-
beled
(h). (b) A z slice which
bisects the radial arms is
shown. Additional domains
are visible including: the cy-
toplasmic and nncleoplasmic
thin rings ( CR and NR), the
radial arms (RA), the inner
spoke ring (ISR),
and the nu-
clear envelope border (M).
The cytoplasmic ring and at-
tached particles
(CP)
com-
bine to form a cresent shape
in this view. (c) A z slice
which bisects the spokes of the detergent-extracted form is shown. Additional domains are labeled including the transporter (T), cytoplasmic
and nncleoplasmic thin rings
(CR
and NR), and connections to the inner spoke ring
(white arrowheads).
Other domains labeled as above.
Note that the inner spoke domain is split into cytoplasmic and nucleoplasmic domains
(small black arrows on le~)
that join at the midline
of the spokes and interact with the inner vertical domains and the central domain. (d) A z slice which bisects the radial arms of the dform
is shown. Note that the membrane density is missing and the inner spoke ring is more rounded.
Fig. 6. End-on and oblique views (at 45°), as viewed from
the cytoplasmic side, emphasize the connectivity of the lu-
menal ring (Fig. 6, a and b). Within the lumenal ring, the
radial arms attach diagonally in an alternating fashion to
the top and bottom of adjacent lumenal spoke domains. The
spoke-ring complex is colored a medium blue while the lu-
menal ring is darker blue. The cut-away surfaces presented
in Fig. 6, c and d are shown at a higher cutoff to minimize
contributions from the nuclear envelope. In these views, the
individual spokes and a dimeric splitting within the inner
spoke ring are visualized.
NPCs Associated with the Nuclear Lamina after
Detergent Extraction
Nuclear envelope ghosts were obtained by TX-100 extraction
of nuclei, extruded into low salt buffers, and subsequently
spread onto grids for rapid freezing. Under these conditions
NPCs remain attached to the disordered lamina (but see
references 2, 3). Areas of a zero tilt/42 ° tilt pair are shown
in Fig. 1, c and d. The position of the tilt axis is indicated
by the white arrow and the lamina is visible as a diffuse net-
work of fine strands located between the NPCs
(black arrow-
heads
labeled "L" in Fig. 1 c). In general, the NPCs demon-
strate a high occupancy of central ring-like transporters
(white arrowheads
in Fig. 1 c). Averages from 284 dform
NPCs after alignment are shown in Fig. 2 b (untilted) and
e (cylindrical average). The spoke-ring complex is more
asymmetric than observed in the best images from
Necturus
(3) and displays the lower point group symmetry C8. The
radial arms are visible between the spokes at high radius but
are characteristically more diffuse than in maps of mform
NPCs (3). In addition, the NPC transporter is visualized in
projection as an averaged"plug" that may encompass a num-
ber of different transport-related configurations (C. W. Akey,
unpublished data; see references 4, 8). Interestingly, the
transporter is visible in the tilted cylindrical average as a
horizontal band of density within the inner bright elliptical
ring, and the cytoplasmic and nucleoplasmic thin rings are
resolved (Fig. 2 e). Numerous attempts were made to obtain
symmetrical (,,o822) specimens from
Xenopus
without suc-
cess. A projection map from a second dform specimen is
shown in Fig. 2 c. The similarity of the two averages suggests
that this configuration is reproducible and similar maps have
Akey and Radernmcher Architecture of the Xenopus Nuclear Pore Complex 7
Figure
6. Surface views of the 3D map of the reform are presented at two different density cutoffs. The spoke-ring assembly is medium
blue, the lumenal ring is dark blue. (a) The NPC is shown as viewed directly down the eightfold rotation axis from the cytoplasmic surface.
The octagonal central channel formed by the inner spoke ring has a diameter of '~440 ,~ in this view. The lumenal ring comprised of
the lumenal spoke (LS) and radial arm (R/I) domains extend radially from the spokes into the lumen of the nnelear envelope. (b) A view
of the pore complex rotated 45 ° about a vertical axis from the view in a is shown. The diagonal bridging of the lumenal domains by the
radial arm dimers is shown clearly. Remnants of the cytoplasmic particles
(CP)
protrude from the cytoplasmic thin ring (CR) of the NPC.
The nucleoplasmic thin ring is also indicated (NR). (c) A cut-away surface view of the mform NPC is shown with the cytoplasmic surface
upwards. This view is shown at a higher cutoff to eliminate weaker density between adjacent spokes that is contributed in part, by the
nuclear envelope. The positions of the cytoplasmic and nucleoplasmic thin rings are indicated as is the lumenal spoke domain. (d) The
same cut-away surface is shown rotated 45 ° towards the viewer to show the gaps between adjacent spokes which delineate the cytoplasmic
ring and the dimeric splitting that occurs in the inner spoke ring.
been obtained from negatively stained NPCs (11, 52). For
comparison, a map of dform NPCs is shown in Fig. 2 f
(n = 284) in which the pore complexes were obtained from
older
Necturus
oocytes. Note that the preservation of twofold
symmetry within the spokes and inner spoke ring is inter-
mediate in character between the
Xenopus
maps and the best
maps from
Necturus
(3).
Representative xy slices taken from the 3D volume of the
dform NPC, equivalent to those presented from the 3D map
of the reform NPC, are shown in Fig. 7. The in-plane sym-
The Journal of Cell Biology, Volume 122, 1993 8
Figure
7. A montage of averaged xy slices from the 3D volume of the detergent-extracted NPC is shown. A surface model viewed from
the side is included in g to orient the viewer. The 3D map is 64 cubed with a 25 ./~ spacing between xy slices along z. (a) Ring of cytoplasmic
particles (CP): average of slices 17-20. (b) Cytoplasmic thin ring with two major domains in each ring "subunit" and a smaller inwards
pointing domain (*). The very top of the transporter is visible in the center: average of slices 22-24. (c) The cytoplasmic region of the
spoke assembly is shown. Note the left handed pinwheeling of the spoke domains. The central transporter is labeled (T): an average of
slices 28-29. (d) The central section through the 3D map of the dform NPC is shown. The radial arms (RA), lumenal spoke domain (LS),
central spoke domain
(CS),
transporter (T), and inner spoke ring
(ISR) are
labeled. This section is an average of slices 33-34. Note the
local two fold symmetry of the lumenal ring about putative spoke and radial arm twofold axes
(white arrows I
and 2), that are repeated
eight times around the cylindrical NPC. This symmetry is broken by a local twisting of the inner spoke ring and central domains which
are inner-connected at this level in the map (see dotted surface). This counterclockwise twist results in the local twofold axes for the central
and inner spoke domains being radially offset between the twofold axes of the lumenal ring system (dotted white arrow is offset by 45 °
for clarity and is flanked on either side by local symmetry axes 1 and 2 due to the eightfold rotational symmetry of the NPC). (e) Nucleoplas-
mic thin ring: average of slices 43-47. The ring displays two distinct domains marked with asterisks (*). (f) The nucleoplasmic region
of the spoke assembly is shown, comparable to c. Note that the pinwheeling of the spoke domains is righthanded, in a direction opposite
to c, as expected for a structure with approximate twofold symmetry. However, the correspondence in features between c andfis not perfect
due to local disordering of the nucleoplasmic half of the NPC in this map: an average of slices 37-38. (g) Surface map of the dform NPC
viewed from the side with the vertical positions of various slices indicated. Major domains of the spoke ring assembly are labeled: cytoplas-
mic particles
(CP),
cytoplasmic ring
(CR),
outer vertical domain (Fo), lumenal spoke domain (LS), radial arm dimer (RA), and nucleoplas-
mic thin ring
(NR).
Figure 8. An individual wedge-shaped spoke from the 3D map of
the dform is shown, in a rotation series. The axis of rotation is verti-
cal and the cytoplasmic surface is at the top of each image. The
direction of rotation is out of the plane of the page towards the
viewer (or clockwise when viewed from the top of the figure, along
the eightfold axis). Portions of the cytoplasmic and nucleoplasmic
thin rings that interact with the spoke are also included. (a) A view
from the center of the NPC showing the inner spoke domain (IS),
inner spoke ring (ISR), the cytoplasmic and nncleoplasmic thin
rings (CR and NR), and the inner/outer vertical domains (Vi/Vo).
(b) In this view, the spoke has been rotated 100" clockwise. Note
the presence of the central hollow that separates the outer vertical
and the lumenal spoke domains (LS) from the central domain,
marked with an asterisk (*). Disorder in the nucleoplasmic half of
the map is manifested by the smaller volume of the inner vertical
domain on the bottom surface of the spoke. (c) View of individual
spoke-ring components from outside the NPC. This view is related
to b by a clockwise rotation of 80*. The lumenal domain is visible
and is tilted up to the left and down to the fight, where it is bounded
by attachments to the radial arms (RA). (d) Final clockwise rotation
of 90 ° to complete the series. The cytoplasmic particle (CP) pro-
trudes from the cytoplasmic ring. b and d present views of opposite
surfaces of the wedge-shaped spoke.
metry of the central slice (Fig. 7 d) is more complicated than
in the mform NPC. Instead of one set of approximate twofold
axes there are two sets of "local" approximate twofold sym-
metry axes rotationally offset by ,o12-15 °. The first set is in-
dicated by the white arrows labeled "1" and "2" in Fig. 7 d
and reflects the approximate twofold symmetry of the lu-
menal domains and radial arms. The position of the corre-
sponding "spoke" twofold symmetry axis in the second set
(offset by 45* for clarity) is shown by the dashed arrow. This
approximate twofold axis bisects the central and inner spoke
domains. The local rotational distortion that gives rise to the
two sets of offset symmetry axes is less marked for the inner
spoke domains located above and below the central xy slice
(Fig. 8 a). Nearly equivalent slices located above and below
the midplane of the spoke complex show the spokes as "pin-
wheels" with opposite handedness, as expected for a struc-
ture with twofold symmetry (Fig. 7, c and f). However, the
correspondence in detail between the slices is only approxi-
mate. In addition, both the cytoplasmic and nucleoplasmic
thin rings have eight paired domains although the density in
the nucleoplasmic rings is weaker (compare Fig. 7, b and e).
Both the positions and shapes of peaks within the cytoplas-
mic rings of the mform and dform NPCs are similar, includ-
ing the small inward pointing regions. Unexpectedly, the cy-
toplasmic particles are better preserved in the dform NPC
(Fig. 7 a), due possibly to a higher occupancy after specimen
preparation. Cylindrical features at the centers of the xy
slices are described in the next section on the NPC trans-
porter.
There are large and intriguing differences present in the
3D maps of the mform and dform NPCs. In comparing
cross-sections from the two maps (Fig. 5, a and c), it appears
that the lumenal domains extend radially to a comparable
distance. Other spoke domains are present in differing orien-
tations relative to their neighboring domains. For example,
the outer rod of vertical density in the mform NPC (Fig. 5
a) is kinked in the dform NPC (Fig. 5 c) rather than straight.
Moreover, two vertical domains (Vo) appear to link the lu-
menal domain to the cytoplasmic and nucleoplasmic rings.
The outer vertical domains also may form lateral contacts
with the inner vertical domains (Vi). The "twofold" related
inner vertical domains connect the inner spoke domain to the
cytoplasmic and nucleoplasmic rings, respectively, although
the lower connection is more disordered in the map of the
dform. The structural role of the central domain may be piv-
otal, as a radial displacement of this domain may possibly
have occurred in response to relative movements of the lu-
menal and outer vertical domains (Fig. 7 d). It seems likely
that mechanical distortions may have induced the observed
rotational displacement of this domain from the spoke two-
fold axis. In addition, the inner spoke domain is a dimer of
two smaller domains that are connected to the central do-
main by a pair of strand-like connections at the midplane of
the map (dotted line in Fig. 7 d). These strandlike connec-
tions may move outwards in the presence of the nuclear enve-
lope and contribute to the ring of density observed at a radius
of 420--430/~ in the mform NPC.
A cross-section located 22.5 ° clockwise from Fig. 5 c, bi-
sects the radial arms as shown in Fig. 5 d. A number of fea-
tures in common with the mform NPC are visible including
the cytoplasmic particles, the cytoplasmic and nucleoplas-
mic rings, the radial arms and the inner spoke ring. However,
the inner spoke ring has become more rounded and the verti-
cal density located between the spokes in the 3D map of the
mform NPC is absent. This latter density is abolished after
detergent extraction; hence, it may originate in part from the
nuclear envelope pore in which the NPC resides. The inner
spoke ring plays a critical role in maintaining the integrity
of the NPC, yet is rather variable when comparing projection
maps of symmetric and asymmetric dform NPCs (this work;
and references 3, 34) with maps of mform NPCs (3, 4). This
variability may reflect differing interactions of the inner
spoke ring with the encircling nuclear envelope, as the result
The Journal of Cell Biology, Volume 122, 1993 10
of detergent extraction or specimen preparation. In addition,
the thin rings in the scaled 3D map of the reform NPC are
6-10% smaller in diameter than in the dform NPC. This sug-
gests that the thin rings may be compressed radially as the
result of an inward-directed pressure associated with the nu-
clear envelope.
Domain connectivity within a wedge-shaped spoke is
shown in a rotation series of four surface maps in Fig. 8, with
the rotation axis aligned vertically. The path of strongest
connectivity may run as follows: the cytoplasmic portion of
the inner spoke domain connects to the ridgelike inner verti-
cal domain, which interacts with both the cytoplasmic thin
ring and the outer vertical domain. The outer vertical do-
main contacts the lumenal domain which is linked to adja-
cent spokes by radial arm dimers forming a lumenal ring.
The connectivity sequence described for the cytoplasmic
half of the spoke is continued around the nucleoplasmic half
as shown in Fig. 8, c and d. The asterisk in Fig. 8 b marks
the location of the central domain which is intimately as-
sociated with the inner spoke domain and rotationaily offset
from the spoke midline. The surface morphologies of the
spoke ring assembly presented in this report and the 3D map
described by Hinshaw and co-workers (34) are similar but
not identical. Importantly, comparisons between maps of
distinct but related forms have allowed us to make a qualita-
tive assessment of domain connectivity and potential flexibil-
ity, after compensating for disorder within the nucleoplas-
mic halves of our 3D maps.
Color coded surface maps of the dform NPC structure are
presented in Fig. 9 (a-f). The color coding is as follows: the
lumenal ring is dark blue, the spoke-ring complex is
medium blue and the transporter is pink. As with the mform
NPC, the radial arm dimers connect lumenal domains in ad-
jacent spokes to form a lumenal ring of nucleoporins (Fig.
9, a and b). Furthermore, the lumenal domain does not pro-
trude from the spoke as strongly but rather is more closely
in contact with the outer vertical domains located above and
below it (Fig. 9, b and c). Eight large particles are associated
with the cytoplasmic ring. Their outermost portions mediate
attachment to the cytoplasmic thin ring (Fig. 9 c) while the
innermost regions are linked by a weak ring of density (Fig.
9, a and b). This ring was not observed previously but may
represent cytoplasmic filaments that have collapsed onto the
NPC as the aqueous buffer layer thins during specimen
blotting.
A surface map of the dform NPC in which the transporter
and half of the spoke-ring complex have been computation-
ally removed is shown in Fig. 9 d. In this view the alternating
sizes of the domains which comprise the inner spoke ring are
visible as are the cytoplasmic and nucleoplasmic thin rings.
The inner spoke domain is tilted 'o7 ° clockwise from the
vertical axis, towards the cytoplasmic surface and is com-
prised of two sub-domains related by an approximate twofold
axis. The inner and outer vertical domains connect the thin
rings to the inner spoke ring and outer portions of the spokes,
respectively. These domains are emphasized by the small
hollow feature that radially separates the lumenal domain
from the inner spoke ring/central domain (Fig. 9,
d-f).
Moreover, the observed rotational displacement of the cen-
tral domain within the spoke may have accentuated the size
of the hollow. This feature was not visualized in negatively
stained NPCs, possibly because specimen compression as-
sociated with drying ('o35-50%) may have closed this small
hollow (see reference 34).
The NPC Transporter: A Tripartite
Cylindrical Assembly
The NPC transporter was not preserved in specimen used to
calculate the 3D map of the mform NPC; therefore, this map
has a large central channel. However, the central transporter
is present at a very high occupancy (~75-80%) in the dform
NPCs. A corresponding ring of density is observed at the
centers of the xy slices within the 3D map (Fig. 7,
b-e), and
depending on the level within the map this ringlike feature
is either hollow or appears to be partially occluded by "en-
dogenous" material. A different perspective is obtained by
viewing the cross-sections shown in Fig. 5, c and d. In these
z-slices, the NPC transporter is centrally tapered giving the
impression of a tripartite structure constructed with very ap-
proximate twofold symmetry. The transporter appears to be
hollow but additional density is present in the nucleoplasmic
half of the map. This axial material may represent an aver-
aged view of substrates caught "in transit" within the trans-
porter (4, 8), coupled with residual noise peaks. The trap-
ping of substrates during enucleation into buffers without
ATP may occur rather readily (3, 8), as transport through the
NPC is probably a highly cooperative process that requires
soluble factors and ATP (1, 44, 46).
The cytoplasmic and nucleoplasmic surfaces of the trans-
porter appear solid; a central channel entrance is not pres-
ent. It cannot be ascertained if this is an intrinsic feature of
the transporter or alternatively, may reflect the average super
positioning of substrates docked at entrances to the central
transport channel (4, 8). The external tapering of the trans-
porter c~linder is mirrored internally. The channel diameter
is ~90A wide at the waist and expands out to ll0A in the
cytoplasmic half and 160/~ in the nucleoplasmic half of
the transporter. To some extent, these differences may reflect
the presence of the axial material in the nucleoplasmic half
of the transporter. The walls of the transporter are thinnest
at the central taper, "o75-90~ These estimates will change
with the collection of improved data on NPCs with defined
configurations, as partial occupancy of substrates within the
transporter may cause the walls to appear thicker after rota-
tional averaging. Interestingly, the small diameter of the
putative transport channel at the central waist would suggest
that this region, located between possible gating assemblies
at either end, may be flexible and undergo considerable ex-
pansion during substrate translocation (17, 61).
In surface views of the NPC, the transporter appears as a
central protruding cylinder with a fiat cytoplasmic surface
(Fig. 9,
a-c).
The complete transporter is revealed as a
tripartite hourglass-shaped cylinder with an axial dimension
of'o625~ in views in which half of the spoke-ring assembly
has been computationaUy cut away (Fig. 9, e and f). The cen-
tral waist of the transporter is "o320/~ in diameter while the
putative gating assemblies located at either end have maxi-
mal diameters of "o420/~. The transporter in this structure
is a global average of at least three to four different configura-
tions that may be transport related (unpublished data);
hence, additional data and classification analysis will be
needed to resolve individual domains or subunits within the
assembly. Connections between the transporter and the inner
Akey and Radermacher Architecture of the Xenopus Nuclear Pore Complex 11
The Journal of Cell Biology, Volume 122, 1993 12
spoke ring form a set of eight inter-connected channels (Fig.
9, a-c). The
general shape of these channels can be seen in
Fig. 9 e, which is shown at a higher density cutoff such that
the connections between the transporter and the spoke-ring
assembly are eliminated. These internal channels were ob-
served previously in projection, in class averages of mform
NPCs with central transporters (4). The channel cross-
sections are roughly the correct size (80-90/~) to allow
diffusion of ions, small molecules, and small nonnuclear
proteins across the pore complex (49). Finally, the NPC-
transporter is vertically displaced by roughly 50A towards
the cytoplasmic surface of the NPC, away from the center of
the spoke assembly (Fig. 9, e and f). This displacement may
have occurred as the cumulative result of mechanical stresses
during isolation; similar forces may have induced the ob-
served asymmetry between the top and bottom halves of the
spokes (see Figs. 8 and 9 e).
We maintain a high degree of confidence in the global
structure of the NPC-transporter. First, 3D maps computed
from half and quarter datasets are similar. Second, the size
of the two globular end domains within the transporter, cou-
pled with the approximate twofold symmetry, suggests that
each transp,,o,,rter may be capable of splitting into "half-
transporters' equivalent to the ,,o360/~ diameter pluglike
particles obtained during disassembly studies (62). Third,
the 3D map suggests that the central, disordered cylindrical
rod of density observed previously in an initial 3D map (62)
may be equivalent to the transporter. Fourth, Np-gold (23)
and mRNPs (28, 40, 61) share a similar topology during
transport, as both substrates are restricted to a central chan-
nel that spans the entire NPC. These data are explainable in
terms of the NPC transporter which may provide a macro-
molecular channel that traverses the entire NPC.
Discussion
A detailed molecular analysis of the structure and function
of the NPC has proven difficult due in part to the size, com-
plexity and low abundance of this intracellular leviathan
(39). However, our knowledge has increased in parallel with
the development of techniques to visualize and quantitatively
analyze the pore complex, in combination with newer
methodologies which are now allowing the biochemical and
genetic identification of constituent nucleoporins. Further-
more, the approach of combining cryo-electron microscopy
and X-ray diffraction analyses with biochemical and genetic
information is proving to be a powerful tool in the functional
dissection of large macromolecular assemblies (35, 60, 63).
As a first step in understanding the dynamics of NPC confor-
mation and function, we have used cryo-electron microscopy
and single particle 3D methods to analyze the structure of
NPCs associated with either the nuclear envelope or
detergent-extracted nuclear ghosts. To a first approximation,
the NPC proper can be described in terms of the spoke-ring
complex and the central transporter (but see Fig. 11). In our
experience, both assemblies may exhibit differing conforma-
tions or configurations in isolated specimen (3, 4) and we
fully expect that some of these changes may be induced by
sample isolation and preparation for EM. However, it is our
working hypothesis that both the apparent conformational
flexibility of the spoke-ring complex (3) and the observed
configurations of the transporter (4, 8) may be relevant in a
fundamental way to processes that occur in vivo.
Conformational Flexibility and a Minimal Domain
Model of the NPC
A comparison of the two 3D structures reported in this work
reveals that the spoke-ring complex may be rather flexible
and allows a qualitative analysis of domain positions and in-
teractions. Idealized domain models for the m- and dforms
ofXenopus
NPCs are shown in Fig. 10, redrawn from density
maps of z-slices cut along the putative "spoke" (Fig. 10, a and
c) and "radial arm" twofold axes (Fig. 10, b and d). Inferred
positions of the nuclear envelope and the NPC transporter
have been indicated in Fig. 10, a and b. The spoke assembly
is twofold symmetric about the midplane of the structure as
shown in the best 2D maps (3, 4, 42, 62) and in a recent 3D
map of negatively stained NPCs (34). Although there are
varying degrees of distortion present in the nucleoplasmic
halves of our 3D maps, the major features support this in-
terpretation. Inspection of the two 3D maps suggests that
conformational changes involving the spoke domains and
thin rings have occurred, possibly as the result of detergent-
extraction, osmotic shock within the lumen of the nuclear
envelope or mechanical stresses during specimen prepa-
ration.
As an aid in visualizing the overall design of the NPC, a
3D ribbon diagram of the membrane-associated NPC is
shown in Fig. 11. In addition, this diagram includes periph-
eral assemblies visualized by other methods that form part
Figure
9. A gallery of color coded 3D surface views is shown of the detergent-extracted NPC. The spoke-ring assembly including cytoplas-
mic particles
(CP)
is medium blue, the lumenal ring is dark blue and the central NPC-transporter is pink. (a) The NPC is shown from
the cytoplasmic surface viewed at a glancing angle of 10 °. Note the attachments between the transporter and the surrounding spoke ring
assembly and the thin ring of density which links up adjacent cytoplasmic particles. The lumenal spoke, radial arm domains and the cyto-
plasmic thin ring are labeled. (b) A view of the NPC rotated 45 ° about a vertical axis is shown. The lumenal ring is highlighted in this
view. The cytoplasmic particles and the nucleoplasmic thin ring are labeled. (c) A view of the NPC from the cytoplasmic surface is shown
in the same orientation as a, but with the top portion of the cytoplasmic particles removed to show the packing of the transporter within
the inner spoke ring. Entrances to possible diffusion channels are delineated by the connections between the transporter and the inner
spoke ring. (d) A cut-away view of the spoke-ring assembly is presented with the central transporter removed to show the internal surface
of the spoke-ring assembly. The cut was made along the approximate spoke twofold axis which bisects two of the wedge-shaped spokes.
The lumenal ring and the cytoplasmic particles are indicated. (e) A second cut-away view is shown with the transporter left intact. In
this view the cutting plane is oriented along a putative radial arm twofold axis, between adjacent spokes. The packing of the transporter
within the wedge-shaped spokes is visible as are opposite surfaces of the spokes (compare with Fig. 8). (f) A cut-away side view of the
NPC is shown in the same orientation as d with the transporter left intact. Internal channels located between the hourglass-shaped trans-
porter and the inner spoke ring are revealed. The transporter spans the width of the NPC and the local twofold axis of the transporter
is offset from the approximate twofold axis of the spoke-ring assembly by ,,~50 A, towards the cytoplasmic/top surface.
Akey and Radermacher
Architecture of the Xenopus Nuclear Pore Complex
13
Figure 10. Diagrammatic cross-sections are shown of the membrane-associated and detergent-extracted NPCs. Peripheral assemblies like
the cytoplasmic filaments and nucleoplasmic cages have been omitted for clarity. The density maps presented in Fig. 5 and other data
were used to deduce a minimal domain model. In addition, this analysis allowed conformational rearrangements involving the lumenal
spoke domain, the outer vertical domains and the central spoke domain to be deduced. The diagrams have been idealized by enforcing
exact twofold symmetry about the membrane plane for the spokes. The cytoplasmic and nucleoplasmic rings are not presumed to be identical
and the central hollow in the spokes has been exaggerated for clarity. The diameter of the spoke ring assembly excluding the radial arms
is '~1,200/~. The domain designations are: inner spoke domain (IS), inner spoke ring (ISR), cytoplasmic and nucleoplasmic thin rings
(CR and NR), inner and outer vertical domains (Vi/Vo), central spoke domain (CS), lumenal spoke domain (LS), transporter (T), putative
cytoplasmic and nucleoplasmic gating assemblies (CG/NG), radial arm dimers (RA), cytoplasmic particles (CP), and inner and outer nu-
clear membranes (INM and ONM). (a and b) Z slices from the mform map are shown, cut along putative twofold axes aligned with the
spokes and the radial arms. The nuclear envelope and NPC transporter are shown superimposed onto their approximate locations (features
shown with dotted lines and lighter shading). Note that the vertical rod of density within the spoke in Fig. 5 a has contributions from
the nuclear envelope, as well as the central and the outer vertical domains. The lumenal spoke domain and the radial arms protrude into
the lumen of the nuclear envelope. This model predicts the existence of more than one type of membrane spanning protein within the
pore complex. Possible positions of the macromolecular transport channel (MTC) and the passive diffusion channels (PC) are indicated
by vertical dashed lines in a and b, respectively. (c and d) Idealized z slices are shown for the 3D map of the dform. The size of the cavity
located between the inner spoke/central spoke domains and the lumenal spoke domain has been exaggerated (h). In addition, connections
between the transporter and the nueleoplasmic face of the inner spoke domains have been included although they are weak in the original
map, due to local disorder. Note the overall change in shape of the inner spoke ring in b and d, which suggests that the nuclear envelope
has a significant effect on the conformation of this structural element.
of the extended model of the NPC (2, 28, 33, 36, 55, 57).
Individual domains have been reduced in size (or in some in-
stances omitted) for clarity; for example the inner spoke ring
has been reduced in height to allow visualization of the cen-
tral NPC transporter. However, this reduction causes a dis-
tortion of the inner vertical domains within the wedge-
shaped spokes. Hence, the ribbon diagram should be used
to orient the various subassemblies of the NPC, the model
cross-sections in Fig. 10 are more accurate in terms of the
spatial relationships between spoke domains. Simply, the ar-
chitecture of the spoke-ring assembly can be described in
terms of four colinear rings, that use homologous and heter-
ologous interactions between domains to maintain the in-
tegrity of individual spokes while mediating assembly of the
eightfold symmetric pore complex. First, the inner spoke
ring is formed by domains that bridge the inner domains
within adjacent spokes to form an eightfold symmetric annu-
lus. The inner spoke ring appears to be the central organizing
element of the spoke-ring complex and may function as an
adaptor between the spoke-ring complex and the NPC trans-
porter, to which it is connected (Fig. 11). Second, the cyto-
plasmic and nucleoplasmic thin rings may stabilize the
spoke-ring assembly by making intimate contacts with the
inner and outer vertical domains in each of the eight spokes.
Third, the anchoring of the spoke assembly within the nu-
clear envelope may be stabilized by the formation of a lu-
menal ring, comprised of eight lumenal domains bridged by
radial arm dimers (Fig. 11). Importantly, the position of the
The Journal of Cell Biology, Volume 122, 1993 14
Figure 11. A 3D representation of the
membrane-associated NPC is presented as
a ribbon diagram. Structural information
from other groups has been incorporated to
give an overall impression of our current
knowledge of the extended architecture of
the NPC (2, 28, 33, 36, 55). For clarity, the
dimensions of certain features have been
minimized and the top cytoplasmic ring
(CR) is sliced open to reveal the interior.
For example, the inner spoke ring (ISR) is
shorter and thinner than in the real structure
(see Fig. 9, d-f). These changes result in
some artistic adjustments being made in the
shape of the inner vertical spoke domains
which link up the ISR to the cytoplasmic
thin ring. The spokes are wedge shaped
with seven domains in the diagram. In actu-
ality the inner spoke domains (pan of the
ISR in this diagram), the central and the lu-
menal domains are dimers, giving a total of
10 domains per spoke. The diagram empha-
sizes the construction of the NPC from four
coUinear ring systems including the cyto-
plasmic and nucleoplasmic thin rings (CR/
NR), the inner spoke ring (comprised of the
inner spoke domains and the material be-
tween spokes) and the lumenal ring which
links up adjacent spokes within the lumenal
space. The lumenal ring is composed of the
lumenal domains (LS) and the radial ann
dimers (RA). Additional components of the
NPC proper are labeled: cytoplasmic parti-
cles (CP), the spokes (S), and the tripartite
transporter (T). The precise attachment
sites of peripheral assemblies are not known. Cytoplasmic components include thin filaments (CF) which attach to the exposed thin ring
and the outer nuclear membrane (ONM). On the nucleoplasmic surface, the lamina is apparently anchored into the inner nuclear membrane
(INM) by a membrane receptor (RL). An octal~onal cage or basketlike structure (NC) is attached to the outer surface of the nucleoplasmic
thin ring. This structure may extend ,,ol,000 A into the nucleoplasm. Moreover, adjacent NPCs may be linked into groups by the nuclear
envelope lattice (NEL) which interconnects distal rings (DR) of the cages (33). Finally, thin bridges of material connect each half of the
transporter to the inner spoke ring, thereby forming a set of interconnected channels that may share a common vestibule formed by the
packing of the centrally tapered transporter within the inner spoke ring.
central domain within a wedge-shaped spoke may be pivotal,
as this domain appears capable of moving ~35--40/~ radially
in a rearrangement that may involve a repositioning of the
outer vertical and lumenal domains (compare a and c in Fig.
10). Overall, the dimeric inner spoke, central, and lumenal
domains each contribute a domain to the vertical inner and
outer domains to form half of a wedge-shaped spoke with a
total of five domains. Subsequently, each half spoke is paired
about a central twofold axis (3, 34, 42) to generate a com-
plete spoke and eight of these units combine with the cyto-
plasmic and nucleoplasmic thin rings to form an articulated
spoke-ring assembly.
A comparison of the 3D map of Xenopus NPCs reported
by Hinshaw and co-workers (34) with our work, reveals that
the basic features of the spoke-ring complexes are remark-
ably similar including a sharing of the same absolute handed-
ness. The 3D maps of both dform NPCs were produced using
the random conical tilt reconstruction method (50), with
different image processing packages (SPIDER/SUPRIM) and
different specimen preparative methods (frozen-hydrated
specimens vs. negative-staining). The similarity of the final
3D structures serves to validate both the 3D reconstruction
methods and the reliability of the resulting maps. Although
rapid freezing can potentially preserve structural order to
atomic resolution, specimen dependent factors generally re-
sult in a much lower attainable resolution. For example, the
nucleoplasmic halves of the 3D maps may be partially disor-
dered resulting from the transduction of mechanical stress
along the fibrous lamina and its attachments to the NPC
(2,3). Alternatively, changes in spoke-ring conformation
may result from hydration changes of the nucleoplasm
(swelling), enucleation, and the possible loss of lumenal
components.
Cytoplasmic Particles and Peripheral Assemblies
Fingerprinting experiments first demonstrated the existence
of asymmetrically shaped particles on the cytoplasmic sur-
face of the NPC with a diameter of ~ 220A (62). Subse-
quently, Reichelt and co-workers (52) measured particles
released from the nuclear envelope with diameters of
230-270A and showed them to have a mass of ~6-6.7 Mda.
These particles were associated with the thin rings in some
instances; however, their mass appears to preclude them
Akey and Radermacher
Architecture of the Xenopus Nuclear Pore Complex 15
from being ribosomes. Particles are found associated with
the surface of the cytoplasmic thin rings in both 3D struc-
tures presented in this work. The particles in the map of the
dform are better preserved and have dimensions of ,,o 230 x
160/~ when viewed along the eightfold axis from the
cytoplasm. The shape of the particles is similar to particle
profiles observed in negatively stained specimens obtained
by fingerprinting (62), suggesting that these are true particles
rather than collapsed filaments (36, 55). At a lower density
cutoff, a thin ring appears to inter-connect the eight cytoplas-
mic particles in the dform NPC. This ring may originate
from the collapse of thin cytoplasmic filaments during ad-
sorption of the specimen to the carbon film (33, 55). Other
peripheral assemblies including the nucleoplasmic cages/
baskets (33, 36, 54, 55, 57), the nuclear envelope lattice (33)
and attachments to the nuclear lamina were not retained dur-
ing specimen preparation or alternatively, were too strongly
disordered to be visualized (see Fig. 11 for the relative posi-
tions of these components).
Interactions with the Nuclear Envelope
The NPC resides within a pore in the nuclear envelope
formed by the fusion of the inner and outer nuclear mem-
branes. In the current 3D map of the mform NPC, the nu-
clear envelope appears to conform to the spoke-ring assem-
bly while spanning exposed lateral surfaces of the spokes
between the two oppositely facing thin rings. In addition, the
ring of density observed in projection maps of_the mform
NPCs, occurs at an average radius of 420-430/~ in the 3D
map and the central spoke domain traverses this feature. Fi-
naUy, the lumenal spoke domains and the radial ann dimers
form a ring of nucleoporins on the lumenal side of the nu-
clear membrane. We suggest that this lumenal ring may play
a role in both the assembly and anchoring of the pore com-
plex. Gp-210 a membrane glycoprotein component of the
NPC with a single transmembrane domain has been pro-
posed to play a role in anchoring the NPC to the nuclear
envelope (32, 64). The 3D map of the mform NPC suggests
that the large lumenal domain of gp-210 may be localized to
either the radial arms or the lumenal domains. In either case,
the possibility exists that the lumenal portion of gp-210 may
mediate the formation of homodimers, although this species
may not be stable when the molecule is removed from the
context of the spoke-ring assembly.
Recently, Greber and Gerace (31) have shown that micro-
injection of total hybridoma mRNA encoding a gp-210
specific IgG (RL27) into tissue culture cells, results in the
synthesis and transport of antibody into the lumen of the ER
and the nuclear envleope, where it binds gp-210. Remark-
ably, the nuclear import of micro-injected nucleoplasmin
was reduced fourfold and diffusion of a 10-kD dextran was
also diminished in these cells. What is the structural basis
for this dramatic down regulation of active transport and
diffusion from the lumenal side? Depending on the location
of the epitope recognized by RL27, the functional effect(s)
induced by antibody binding would have to be propagated
over a radial distance of * 350-520A, to impinge on the
putative diffusion and macromoleeular transport channels.
Although of non-physiological origin, the plasticity of the
spoke-ring assembly observed in our maps may mimic cer-
tain aspects of a possible "conformational signaling" mecha-
nism in which nucleocytoplasmic transport is globally regu-
lated by signals originating from the lumenal side of the pore
complex. For example, it is conceivable that RL27 binding
to the lumenal ring may trigger rearrangements within the
outer vertical and lumenal domains, which in turn are trans-
mitted to the inner spoke ring by radial displacement of the
central spoke domains. Of course, alternative routes are pos-
sible. However, the inner spoke ring may play a critical role
in this scenario, as it makes connections to the NPC-
transporter that define an internal network of channels.
Therefore, it is conceivable that changes within the spoke-
ring complex may be transmitted to putative gating assem-
blies within the central transporter by way of the inner spoke
ring. Overall, our observations on the conformational
plasticity of the spoke-ring complex in vitro, lend support
to the emerging hypothesis that the spokes may transduce
signals provided by as yet unidentified lumenal or intracellu-
lar effeetors, thereby modulating global transport properties
of the NPC under physiological conditions (22, 31).
Internal Eh'ffusion Channels
Micro-injection studies with dextrans (49) and polyvinyl-
pyrrolidone-coated gold particles (20) have established that
diffusion channels with a limiting diameter of ~90-100/~
may be present in the pore complex, resulting in an estimated
transport cutoff for non-nuclear proteins and small mole-
cules of •40 kD. The 3D map of the dform NPC suggests
that 8 diffusion channels may be located internally between
the inner spoke ring and the central NPC transporter, with
16 entrances (eight from either side) delineated by connec-
tions between the transporter and the inner spoke domains
(Fig. 10, a and b). The 16 entrances may link up in a common
cylindrical vestibule created by the packing of the centrally
tapered transporter within the surrounding inner spoke ring.
A clearer picture of this internal channel network should
emerge with higher resolution data on a defined transporter
configuration. Recently, eight possible diffusion channels
have been hypothesized by Hinshaw et al. (34) to occur be-
tween adjacent spokes at a radius outside the inner spoke
ring. Our current 3D map of the reform NPC in vitro does
not directly support this proposal. However, it is conceivable
that the nuclear membrane may bulge outwards between ad-
jacent spokes in vivo while remaining tethered to the central
spoke domains, thereby creating "peripheral" diffusion chan-
nels for soluble and membrane proteins between the spokes.
Additional experiments are needed to determine the radial
position(s) of diffusion channels within the NPC in vivo.
The NPC Transporter and Models of the
Translocation Step
Previous work in projection has established that a central
ringlike structure is present within the NPC. This assembly
has been named the transporter, as it appears to define the
macromolecular transport channel for NP-gold conjugates
and endogenous substrates (3, 4, 8). In this work, the 3D
map of the dform NPC demonstrates a central cylindrical
feature that may represent a "global" average of transporters
trapped in a number of transport-related configurations dur-
ing enucleation. Interestingly, the averaged transporter has
a tripartite structure and is comprised of two "globular" as-
semblies which join face-to-face about an approximate two-
fold axis to form a cylinder with a tapered center. The NPC
The Journal of Cell Biology, Volume 122, 1993 16
transporter is hollow with 75-90/~-thick walls and a total
length of 625/k. Studies of the nuclear import of karyophilic
proteins suggest that nuclear import occurs within a central
channel of the NPC as follows. Initially, microinjected sub-
strates accumulate at the nuclear periphery in a temperature
independent fashion, ATP-dependent vectorial translocation
then ensues after substrate binding to the NPC (46, 47, 53).
Further mapping studies using nucleoplasmin-gold have
suggested that transport involves substrate docking to the
central NPC transporter, before a possible gating event
within this assembly (8).
A number of models for the translocation step in nucleo-
cytoplasmic transport have been discussed. For instance, it
has been suggested that substrates might move along illa-
ments postulated to traverse the NPC aided by an ATPase/
motor molecule (10). Recent data have confirmed observa-
tions that filaments are attached to both the cytoplasmic and
nucleoplasmic surfaces of the NPC (33, 36, 53, 55, 57).
However, our maps suggest that there is no physical room for
filaments to traverse the NPC itself when the transporter is
present; hence, this mode of translocation would appear un-
likely. A second model has posited that a substrate bearing
container may transit through the center of the spoke assem-
bly, thereby providing protection for mRNPs. This model
was recently brought to the forefront with the suggestion that
vaults, a large assembly with 822 symmetry, might be in-
volved in transport (37). However, images in which large
mRNPs (28, 40, 61) and Np-gold (23) are caught transiting
the entire width of the NPC would argue against the transport
container hypothesis. Moreover, a container mechanism ap-
pears unnecessary as many mRNPs spend a large fraction of
their lifetimes associated with fibrous tracts (reviewed in ref-
erence 38). Finally, the dimensions of the vaults and the NPC
transporter in frozen-hydrated specimens are dissimilar
(vaults: 490 x 275]t; transporter: 625 x 420/~) and their
morphologies are rather different.
The recent visualization of octagonally arranged baskets
or cages on the nucleoplasmic face of the NPC has sparked
interest in the possible roles of these structures in transport
(33, 36, 54, 55). However, the strict localization of the cages
on the nucleoplasmic side of the NPC would pose a topologi-
cal problem to possible roles in a bidirectional gating mecha-
nism. In addition, recent micrographs of the nucleoplasmic
surface of metal-coated amphibian oocytes have demon-
strated the presence of a filamentous lattice that intercon-
nects adjacent cages at their distal rings (33). Perhaps, the
cages provide access to the nucleoplasmic face of the NPC
transporter for macromolecules within the context of the
high density of peripheral chromatin and the thickened lam-
ina present in many cells.
Labeling studies with WGA and MAb414 suggest that the
transporter may be eightfold symmetric, as these bivalent
probes often label individual transporters in groups of four
(8, 57). A direct determination of the symmetry of the trans-
porter must await monospecific FAbs for components of the
transporter. Recently, a macromolecular lock/double iris hy-
pothesis has been proposed to describe the translocation of
substrates catalyzed by the NPC (4). In this model the trans-
porter is proposed to have ,~822 symmetry (reviewed in
references 5 and 6). The limited number of NPCs in our
dataset coupled with the apparent admixture of differing
transporter configurations, results in a globally averaged
view of the transporter that does not allow a direct confirma-
tion of the irislike gating mechanism at this stage. However,
the tripartite morphology of the NPC transporter when com-
bined with data on substrate translocation through the
centers of NPCs (4, 8, 23), lends support to the hypothesis
that this process may involve two oppositely facing gates. In
this model, the two globular end domains of the NPC trans-
porter would correspond to possible gating assemblies. It is
suggested that these putative gates may open asynchronously
to regulate transport along a central channel like a macro-
molecular lock (4, 7). Furthermore, it is tempting to specu-
late that the translocation mechanism may use induced
changes in channel diameter, within the centrally tapered re-
gion of the transporter, to communicate the status of oppo-
sitely facing gates.
Conclusions
The NPC is comprised of a symmetrical or pseudo-sym-
metrical spoke-ring complex located in the plane of the nu-
clear envelope, that frames a central transport machine. In
addition, data from other groups indicates that the NPC
makes peripheral attachments to cytoplasmic filaments, par-
ticles, nucleoplasmic cages/baskets and the lamina to form
an extended transport assembly (see Fig. 11; reviewed in
references 7 and 55). Our 3D maps suggest that each half
spoke is comprised of five domains that appear capable of
moving relative to each other, providing an inherent confor-
mational flexibility within the spoke-ring assembly. There-
fore, it is hypothesized that the structural plasticity observed
in vitro may play a role in the global regulation of transport
in vivo, at the level of the NPC. In addition, the framework
of the pore complex is firmly anchored within the nuclear
envelope as the tip of each spoke penetrates into the nuclear
lumen. In this way, a lumenal ring of nucleoporins is formed
by the association of adjacent lumenal spoke domains with
intervening radial arm dimers. Interestingly, the nuclear
envelope appears to conform to the surface of the octagonal
spoke complex while leaving the thin rings exposed to their
respective cellular compartments (also see reference 33).
However, the lumen between nuclear membranes may un-
dergo osmotic swelling during the isolation of macro-nuclei.
Hence, it is not known whether the membrane-associated
NPC described in this work represents the in vivo conforma-
tion or merely a single structure in a range of possible struc-
tures (reference 34; and this work). Moreover, the visualiza-
tion of the central transporter as an elongated centrally
tapered cylinder, lends support to the proposed macro-
molecular lock hypothesis. In this model, it is suggested that
substrate passage across the nuclear envelope may require
two asynchronous gating events within the context of a cen-
tral channel (4). High resolution data are now needed on dis-
tinct conformations of the spoke-ring complex and the as-
sociated transporter to extend our understanding of this
unique transport machine.
This work has benefited from the suggestions of M. Rout and the support
ofN. Unwin. The computational biology was greatly aided by A. Leith who
made available upgrades of SPIDER/WEB and D. Atkinson, who set up our
computer network. We gratefully acknowledge the advice provided by R.
Nolte on color slide and print making. Finally, we thank colleagues in the
Department of Cell Biology, Stanford University School of Medicine and
the Structural Studies Division, Medical Research Council Laboratory of
Akey and Radermacher
Architecture of the Xenopus Nuclear Pore Complex 17
Molecular Biology, for access to and support of the Philips EM400 series
microscopes.
This work was supported by the National Institutes of Health (R01
GM45377 and R01 GM29169).
Received for publication 22 September 1992 and in revised form 6 April
1993.
ReferGnc£$
1. Adam, S. A., and L. C-erace. 1991. Cytosolic proteins that specifically bind
nuclear location sequences are receptors for nuclear import. Cell.
66:837-847.
2. Aebi, U., J. Cohn, L. Buhle, and L. Gerace. 1986. The nuclear lamina
is a meshwork of intermediate-type filaments. Nature (Lond.). 323:
560-564.
3. Akey, C. W. 1989. Interactions and structure of the Nuclear Pore Complex
revealed by cryo-electron microscopy. J. Cell Biol. 109:955-970.
4. Akey, C. W,
1990.
Visualization of transport-related configurations of the
nuclear pore transporter.
Biophys. J.
58:341-355.
5. Akey, C. W. 1991. Probing the structure and function of the nuclear pore
complex. Semin. Cell Biol. 2:167-177.
6. Akey, C. W. 1992a. The Nuclear Pore Complex. Curr. Opin. Struct. Biol.
2:258-263.
7. Akey, C. W. 1992b. The nuclear pore complex: a macromelecular trans-
porter. In Nuclear Trafficking. C. Feldherr, editor. Academic Press, New
York. 31-70.
8. Akey, C. W., and D. S. Goldfarb. 1989. Protein import through the nu-
clear pore complex is a multi-step process. J. Cell Biol. 109:971-982.
9. Bataille, N., T. Helsen, and H. M. Fried. 1990. Cytoplasmic transport of
ribosomal subunits microinjected into the Xenopus laevis oocyte nucleus:
a generalized facilitated process. J. Cell Biol. 111:1571-1582.
10. lkrrios, M., P. A. Fisher, and E. C. Matz. 1991. Localization of a myosin
heavy chain-like polypeptide to Drosophila Nuclear Pore Complexes.
Proc. Natl. Acad. Sci. 88:219-223.
11. Buhle, L., and U. Aebi. 1985. Correlation of surface topography of metal
shadowed specimens with their negatively stained reconstructions.
Ultramicroscopy. 16:436--450.
12. Carazo, M. J., and J. L. Carrascosa. 1987. Information recovery in miss-
ing angular data cases: an approach by the convex projections method in
three-dimensions. J. Microscopy. 145:23-43.
13. Dabanvalle, M-C., B. Sehulz, U. Seheer, and R. Peters. 1988. Inhibition
of nuclear accumulation of karyophilic proteins in
living
cells by microin-
jection of the lectin wheat germ agglutinin. F_.zp. Cell Res. 174:291-296.
14. Dargemont, C., and L. C. Kulm. 1992. Export of mRNA from microin-
jected nuclei of Xenopus laevis oocytes. J. Cell Biol. 118:1-9.
15. Dingwall, C., and R. A. Laskey. 1986. Protein import into the cell nucleus.
Ann. Rev. Cell Biol. 2:367-390.
16. Dubochet, J., M. Adrian, J. Chang, J. C. Homo, J. Lepauh, A. W.
McDowall, and P. Sbultz. 1988. Cryo-electron microscopy of vitrified
specimens. Q. Rev. Biophys. 21:129-228.
17. Dworetzky, S. I., and C. M. Feldherr. 1988. Translocation of RNA-coated
gold particles through the nuclear pores of cocytes. J. Cell Biol.
106:575-584.
18. Faberge, A. C. 1973. Direct demonstration of eighffold symmetry in nu-
clear pores. Z. Zellforsch Mikrask. Anat. 136:183-190.
19. Featberstone, C., M. Darby, and L. Gerace. 1988. A monoclonal antibody
against the nuclear pore complex inhibits nucleocytopiasmic transport of
protein and RNA in vivo. J. Cell Biol. 107:1289-1297.
20. Feldherr, C. M. 1962. The nuclear annuli as pathways for nucleocytoplas-
mic exchanges. J. Cell Biol. 14:65-72.
21. Feldberr, C. M. 1992. Overview of signal-mediated transport. In Nuclear
Trafficking, C. M. Feldherr, editor. Academic Press, New York. 15-28.
22. Feldberr. C. M., and D. Akin. 1990. The permeability of the nuclear enve-
lope in dividing and nondividing cell cultures. J. Cell Biol. 111:1-8.
23. Feldherr, C. M., E. Kallenback, and N. Schultz. 1984. Movement of a
karyophilic protein through the nuclear pores of oocytes. J. Cell Biol.
99:2216-2222.
24. Fininy, D. R., D. D. Newmeyer, T. M. Price, and D. J. Forbes. 1987.
Inhibition of in vitro nuclear transport by a lactin that binds to nuclear
pores. J. Cell BioL 104:189-200.
25. Forbes, D. J. 1992. Structure and function of the nuclear pore. Ann. Rev.
Cell Biol. 8: 495-527.
26. Frank, J., B. Shimkin~ and H. Dowse. 1981a. SPIDER-a modular soft'ware
system for electron image processing. Ultramicroscopy. 6:343-358.
27. Frank, J., A. Verschoor, and M. Boublik. 1981b. Computer averaging of
electron micrographs of 40S ribosomal subunits. Science (Wash. DC).
214:1353-1355.
28. Franke, W. W., and U. Scheer. 1970. The ultrastructure of the nuclear
envelope of
amphibian
cocytes, a
reinvestigation
I: the mature
oocyte.
J. Ultrastruc. Res. 30:288-316.
29. Gerace, L., and B. Burke. 1988. Functional organization of the nuclear
envelope. Ann. Rev. Cell Biol. 4:335-374.
30. Goldfarb, D. S., J. Gariapy, G. Schoolnik, and R. D. Kornberg. 1986.
Synthetic peptides as nuclear localization signals. Nature (Lond.).
326:641--644.
31. C-reber, U. F., and L. Gerace. 1992. Nuclear protein import is inhibited
by an antibody to a lumenal epitope of a nuclear pore complex glycopro-
rein. J. Cell Biol. 116:15-30.
32. Greber, U. F., A. Senior, and L. Gerace. 1990. A major glycoprotein of
the nuclear pore complex is a membrane-spanning polypoptide with a
large lumenal domain and a small cytoplasmic tail. EMBO (Eur. Mol.
Biol. Organ.)J. 9:1495-1502.
33. Goldberg, M. W., and T. D. Allen. 1992. High Resolution scanning elec-
tron microscopy of the nuclear envelope: demonstration of a new, regu-
lar, fibrous lattice attached to the baskets of the nucleoplasmic face of the
nuclear pores. J. Cell Biol. 119:1429-1440.
34. Hinshaw, J. E., B. O. Carragher, and R. A. Milligan. 1992. Architecture
and design of the nuclear pore complex. Cell. 69:1133-1141.
35. Holmes, K. C., andW. Kabsch. 1991. Muscle proteins: actin. Curr. Opin.
Struct. Biol. 1:270-280.
36. Jarnik, M., and U. Aebi. 1991. Toward a mere complete 3-D structure of
the nuclear pore complex. J. Struct. Biol. 107:291-308.
37. Kedersha, N. L., J. E. Heuser, D. C. Chugani, and L. H. Rome. 1991.
Vaults HI. Vault ribonucleoprotein particles open into flower-like struc-
tures with octagonal symmetry. J. Cell Biol. 112:225-235.
38. Maquat, L. E. 1991. Nuclear mRNA export. Curt. Opin. Cell Biol.
3:1004-1012.
39. Maul, G. 1977. The nuclear and cytoplasmic pore complex: structure, dy-
namics, distribution and evolution. Int. Rev. Cytoi. Suppl. 6:75-186.
40. Mehlin, H., B. Daneholt, and U. Skoglund. 1992. Translocation of a
specific premessenger ribonucleoprntein particle through the nuclear pore
studied with electron microscope tomography. Cell. 69:605-613.
41. Michnud, N., and D. S. Goldfarb. 1992. Microinjected U snRNAs are im-
ported to oocyte nuclei via the nuclear pore complex by three distinguish-
able targeting pathways. J. Cell Biol. 116:851-861.
42. Milligan, R. A. 1986. A structural model for the nuclear pore complex.
In Nucleocytopiasmic Transport. R. Peters and M. Trendelenburg, edi-
tors. Springer-Veriag New York Inc., New York. 111-122.
43. Moll, T., G. Tebb, U. Surana, H. Robitsch, and K. Nnsmyth. 1991. The
role of phosphorylation and the CDC28 protein kinase in cell cycle-
regulated nuclear import of the S. Cerevisiae transcription factor SWI5.
Cell. 66:743-758.
44. Moore, M. S., and G. Blobel. 1992. The two steps of nuclear import, tar-
geting to the nuclear envelope and translocafion through the nuclear pore,
require different cytosolic factors. Cell. 69:939-950.
45. Newmeyer, D. D., and D. J. Forbes. 1990. An N-ethylmaleimide sensitive
cytosolic factor necessary for nuclear protein import: requirement in
sigual-mediated binding to the nuclear pore. J. Cell Biol. 110:547-557.
46. Newmeyer, D. D., J. M. Lucocq, T. R. Burglin, and E. M. DeRobeRis.
1986. Assembly in vitro of nuclei active in nuclear protein transport: ATP
is required for nucleoplasmic accumulation. EMBO (Fur. Mol. BioL or-
&an.) J. 5:501-510.
47. Newmeyer, D. D., and D. J. Forbes. 1988. Nuclear import can be sepa-
rated into distinct steps in vitro: nuclear pore binding and translocation.
Cell. 52: 641-653.
48. Nigg, E. A., P. A. Baeuerle, and R. Luhrmann. 1991. Nuclear import-
export: in search of signals and mechanisms. Cell. 66:15-22.
49. Paine, P. L., L. C. Moore, and S. B. Horowitz. 1975. Nuclear envelope
pe~ility. Nature (Lond. ). 254:109-114.
50. Radermacher, M. 1988. Three dimensional reconstruction of single parti-
cles from random conical tilt series. J. Electron Microsc. Tech.
9:359-394.
51. Radermacher, M. 1992. The structure of the 50S ribosomal subunit from
IF.. coil in frozen hydrated preparation reconstructed with SECRET. Proc.
Xth Eur. Congr. Electron Microsc. In press.
52. Reicbelt, R., A. Holzenburg, E. L. Buhle Jr., M. Jamik, A. Englel, and
U. Aebi. 1990. Correlation between structure and mass distribution of
the nuclear pore complex and of distinct pore complex components. J.
Cell Biol. 110:883-894.
53. Richardson, W. D., A. D. Mills, S. M. Dilworth, R. A. Laskey, and C.
Dingwall. 1988. Nuclear protein migration involves two steps: rapid
binding at the nuclear envelope, followed by slower translocation through
nuclear pores. Cell. 52:655--664.
54. Ris, H. 1990. Application of low voltage high resolution SEM in the study
of complex intracellular structures. Proc. XII Int. Congr. Electron Mi-
croscopy. San Francisco Press, Inc., San Francisco. 18-19.
55. Ris, H. 1991. The three-dimensional structure of the Nuclear Pore Com-
plex as seen by high voltage electron microscopy and high resolution low
voltage scanning electron microscopy. EMSA Bulletin. 21:1.54-56.
56. Robbins, R., S. Dflworth, R. A. Laskey, and C. Dingwail. 1991. Two in-
terdependant basic domains in the nuclooplasmin nuclear targeting se-
quence: identification of a new and distinct class of nuclear targeting se-
quence. Cell. 64:615-623.
57. Scheer, U., M-C. Dabauvalle, H. Merkert, and R. Benevente. 1988. The
nuclear envelope and the organiT~tion of the pore complexes. In
Nucleocytoplasmic Transport. R. Peters, editor. Academic Press, Lon-
don. 5-25.
The Journal of Cell Biology, Volume 122, 1993 18
58. Steme-Marr, R., J. M. Blevitt, and L. Gerace. 1992. O-linked glycopro-
teins of the nuclear pore complex interact with a cytosolic factor required
for nuclear protein import. J. Cell Biol. 116:271-280.
59. Stewart, M., and S. Whytock. 1988. The structure and interactions of com-
ponents of nuclear envelopes from Xenopus oocyte germinal vesicles ob-
served by heavy metal shadowing. J. Cell Sci. 90:409-423.
60. Stewart, P. L., R. M. Bumett, M. Cyrklaff, andS. D. Fuller. 1991. Image
reconstruction reveals the complex molecular organization of adenovirus.
Cell. 67:145-154.
61. Stevens, B. J., and H. Swift. 1966. RNA transport from nucleus to
cytoplasm in Chironomus salivary glands. J. Cell Biol. 31:55-77.
62. Unwin, P. N. T., and R. A. Milligan. 1982. A large particle associated
with the perimeter of the nuclear pore complex. J. Cell Biol. 93:63-75.
63. Wang, G., C. Porta, Z. Chen, T. S. Baker, and J. E. Johnson. 1992.
Identification of a Fab footprint site on an icosahedral virus by cryoelec-
tron microscopy and x-ray crystallography. Nature (Lond.). 355:275-
278.
64. Wozniak, R. W., E. Bartnik, and G. Blobel. 1989. Primary structure anal-
ysis of an integral membrane glycoprotein of the nuclear pore. J. Cell
Biol. 108:2083-2092.
65. Zasloff, M. 1983. t-RNA transport from the nucleus in a eukaryotic cell:
a carrier mediated translocation process. Proc. Natl. Acad. Sci.
80:6436-6440.
Akey and Radermacher Architecture of the Xenopua Nuclear Pore Complex 19
