www.sciencemag.org/cgi/content/full/science.
aaa4136
/DC1
Supplementary
Material
s for
Architecture of the nuclear pore complex coat
Tobias
Stuwe
,
Ana R.
Correia
,
Daniel H.
Lin
,
Marcin
Paduch
,
Vincent T.
Lu
,
Anthony A.
Kossiakoff
,
André
Hoelz
*
*Corresponding author
. E-mail: hoelz@caltech.edu
Published
12 February
201
5 on
Science
Express
DOI:
10.1126/science.
aaa4136
This PDF file includes:
Materials and Methods
Figs. S1 to S13
Tables S1 and S2
References (
10
,
18– 28
)
Cap
tions for
Movies S1 to S4
Other
Supplementary Materials
for this manuscript include the following
:
(available at
www.sciencemag.org/cgi/content/full/science.
aaa4136
/DC1
)
Movies S1 to S4
2
Materials and Methods
Protein expression and purification.
DNA fragments encoding full-length Sec13,
Nup145C (residues 75-712), full-length Seh1, and Nup85 (44-744) were PCR amplified
and cloned as pairs into the pET-Duet1 expression vector (Novagen). Nup145C and
Nup85 were cloned into the first multiple cloning site using BamHI and NotI restriction
sites, whereas Sec13 and Seh1 were cloned into the second multiple cloning site using
NdeI and XhoI restriction sites. Nup84 (residues 1-451) was cloned into a modified
pET28a vector, which contains an N-terminal hexahistidine tag followed by a PreScission
protease cleavage site, using NdeI and NotI restriction sites
(
18
). The expression
construct for Nup120 was described previously (
2
). The selected synthetic antibody
(sAB) fragments of sAB-57
and sAB-87 were cloned into the pSFV4 vector (Peter
Loppnau, Structural Genomics Consortium, University of Toronto) using the restriction
sites NcoI and SalI, and subsequently digested using SalI and BsaI and religated to obtain
the C-terminal hexahistidine tag. The details of the bacterial expression constructs are
listed in table S1.
All proteins were expressed in
Escherichia coli
BL21-Codon-Plus (DE3)-RIL cells
(Stratagene) in Luria-Bertani media. Seleno-L-methionine-labeled (SeMet)
'42 N"D? '47 N"D?0=3"D?
NTD
were produced in a synthetic medium that
suppresses methionine biosynthesis, following standard protocols. For
all nucleoporins,
expression was induced at an OD600 of 0.8 with
0.5 mM isopropyl-
β
-D-
thiogalactopyranoside (IPTG), followed by growth at 18 °C for 18 hours. Cells were
harvested by centrifugation and resuspended in a buffer containing 20 mM TRIS (pH
8.0), 500 mM NaCl, 20 mM imidazole, 4 mM
β
-mercaptoethanol (
β
-ME), and complete
EDTA-free protease inhibitor mixture (Roche).
For purification, cells were lysed with a cell disruptor (Avestin) and DNase I
(Roche) was added to the lysate before centrifugation at 30,000
×
g for 1 hour. The
supernatant was filtered through a 0.45-
μ
m filter (Millipore) and loaded onto a nickel-
nitrilotriacetic acid (Ni-NTA) column (Qiagen) equilibrated in 20 mM TRIS (pH 8.0),
500 mM NaCl, 20 mM imidazole, and 4 mM
β
-ME. Protein was
eluted with
a linear
gradient of 20 mM TRIS (pH 8.0), 500 mM NaCl, 500 mM imidazole, and 4 mM
β
-ME.
Protein-containing fractions were pooled, incubated with either PreScission (GE
Healthcare) or ULP1 protease, and dialyzed overnight at 4 °C against a buffer containing
20 mM TRIS (pH 8.0), 100 mM NaCl, and 5 mM DTT. Next the protein was loaded onto
a Mono Q 10/100 GL ion-exchange column (GE Healthcare) equilibrated in a buffer
containing 20
mM TRIS (pH 8.0), 100 mM NaCl, and 5 mM DTT and eluted using a
NaCl
gradient. Protein-containing fractions were concentrated in a centrifugal filter
(Millipore) and loaded onto a HiLoad Superdex 200 16/60 gel filtration column (GE
Healthcare) equilibrated in a buffer containing 20 mM TRIS (pH 8.0), 100 mM NaCl,
and 5 mM DTT. Protein-containing fractions were pooled and concentrated to 20 mg/mL
for biochemical interaction experiments and CNC reconstitution.
sABs were expressed, harvested, and lysed in a similar fashion, but the cells were
induced at an OD600 of 0.9 with 0.25 mM IPTG and grown at 25 °C for 18 hours. After
lysis, the lysate was incubated at 65 °C for 30 minutes and then cooled on ice for 15
minutes before centrifugation. Protein-containing fractions from the Ni-NTA affinity
purification were pooled and loaded onto a 5 mL HiTrap MabSelect SuRe column (GE
3
Healthcare) equilibrated in
a buffer containing
20 mM TRIS (pH 8.0), 500 mM NaCl, 20
mM imidazole, and 4 mM
β
-ME. The protein was eluted using a linear gradient of
an
elution buffer, containing 0.1 M sodium citrate (pH 3.2). To rapidly increase the pH after
elution, the fractions were collected into tubes containing 200
μ
l of 1 M TRIS (pH 9.0).
The eluted fractions were dialysed against a buffer containing 20 mM TRIS (pH 8.0) and
100 mM NaCl and concentrated to 10 mg/ml for biochemical interaction experiments and
crystallization.
Reconstitution of CNC complexes.
'47 N"D?N"D? (A8<4A F0B?DA858431H2>
-
;HB8B>524;;B4G?A4BB8=6'47 N"D?>A"D? 5>;;>F8=6C74?A>C>2>;34B2A814301>E4
For the
A42>=BC8CDC8>=>5'42 N"D? N"D?
NTD
(A8<4A ?DA85843'42 N"D?
was mixed with a 1.2 fold molar excess of Nup84
NTD
, incubated on ice for 30 minutes,
and loaded onto a HiLoad Superdex 200 16/60 gel filtration column equilibrated in a
buffer containing 20 mM TRIS (pH 8.0), 100 mM NaCl, and 5 mM DTT. Trimer1 and
Trimer2 containing fractions were pooled, concentrated, mixed with a 1.2 molar excess
of Trimer2, incubated on ice for 1 hour, and injected onto a HiLoad Superdex 200 16/60
gel filtration column equilibrated in the same buffer. Fractions containing the
reconstituted CNC were pooled and concentrated to 10 mg/ml for sAB interaction
experiments. For SeMet labeled CNC, SeMet-
;014;43'47 N"D?'42 N"D? 0=3
Nup84
NTD
were purified and used instead of the native proteins. SeMet labeling of
Nup120 rendered the protein insoluble in our bacterial expression system and thus native
Nup120 was used for the reconstitution of the SeMet-labeled CNC. For the generation of
"NB2><?;4G4B=0C8E4>A'4
Met-labeled CNC were mixed with 1.5 fold molar
excess of sAB-57 or a 1:1 mixture of sAB-57 and sAB-87 and loaded onto a HiLoad
Superdex 200 16/60 gel filtration column (GE Healthcare) equilibrated in a buffer
containing 20 mM TRIS (pH 8.0), 100 mM NaCl, and 5 mM DTT. DTT was included in
the buffer as it was necessary for CNC stability and had no effect on the integrity of the
sABs. Fractions containing the various CNC complexes were pooled and concentrated to
10 mg/mL for crystallization.
sAB
selection and characterization.
The generation and screening of conformation
-
specific sABs has been described previously (
5
). Briefly, a modified yeast CNC was
reconstituted with a Nup84
NTD
variant that harbored an N-terminal avi-tag. The complex
was biotinylated in a 2 mL reaction by incubating 40
μ
M protein with a buffer containing
50 mM BICINE (pH 8.3), 100 mM biotin, 10 mM ATP, 10 mM magnesium acetate, and
30
μ
g biotin ligase (BirA) at 30 °C for 2 hours. After labeling, protein was buffer
exchanged using a 5 mL HiTrap Desalting column (GE Healthcare) equilibrated with a
buffer containing 20 mM TRIS (pH 8.0), 100 mM NaCl, and 5 mM DTT and purified
again using a HiLoad Superdex 200 16/60 gel filtration column equilibrated in the same
buffer. The extent of Nup84
NTD
biotinylation and efficiency of capture were tested by
incubating 25
μ
g of protein with 50
μ
L of Streptavidin MagneSphere particles (Thermo
Scientific), washing once with 50
μ
L of a buffer containing 20 mM TRIS (pH 8.0), 100
mM NaCl, and 5
mM DTT, and resolving the bound proteins on a SDS-PAGE gel. Four
rounds of competitive selection were performed using 100 nM (round 1), 50 nM (round
2), 10 nM (round 3), and 10 nM (round 4) biotinylated protein target
and a phage display
library according to previously published protocols (
5
).
In case of sAB-57 biotinylated
4
yeast CNC was used and to eliminate sABs that recognized unassociated CNC
components, 1
μ
M of non-
18>C8=H;0C43 " BD1D=8CB '47 N"D? "D?
NTD
,
'42 N"D? 0=3"D?
NTD
) were used as competitors in all solutions during the last
three rounds of selection. Phages were preincubated with competitors for 1 hour at room
temperature. sAB-
87 was obtained in a selection where biotinylated Nup120
NTD
was used
and no competition was performed.
After successful selection, the specificity of
candidate sABs was tested against the assembled biotinylated yeast CNC, as well as
individual biotinylated subunits using a single point competitive ELISA assay
(
5
). Only
sequence-unique sABs with the desired binding properties were nominated for further
biochemical characterization. To evaluate the binding affinity and specificity of the
selected sABs, 1.5-fold molar excess of sAB was incubated with the reconstituted CNC
or individual CNC components and loaded onto a MonoQ 5/50 GL ion-exchange column
(GE Healthcare) equilibrated in a buffer containing 20 mM TRIS (pH 8.0), 100 mM
NaCl, and 5 mM DTT and eluted using a NaCl gradient. Interacting sABs eluted with the
CNC components, whereas non-interacting sABs eluted prior to the gradient step.
Initially, only sABs that specifically interacted with the fully assembled CNC were
systematically tested in crystallization trials. To improve the diffraction properties of the
"NB
-57 crystals, additional sABs with the ability to bind individual CNC
components were systematically screened for crystal formation. The addition of sAB
-87
C> C74 "NB
-57 complex yielded a new crystal form with distinct packing and
different space group.
Protein crystallization,
heavy metal derivatization
and data collection.
Protein
crystallization was carried out at 21 °C in hanging drops consisting of 1.0
μ
L protein
solution (2 mg/ml) and 1.0
μ
L reservoir solution. Crystals appeared in the monoclinic
space group C2 with one co
?H>5C74"NB
-57 complex in the asymmetric unit. The
crystals were improved by microseeding, which resulted in crystals that grew as thin
plates with maximum dimensions of ~30
×
300
×
300
μ
m
3
within 1 week. Crystals used for
diffraction experiments were grown in 0.1 M MES, pH 6.7, 5 % (w/v) PEG 20000, and
3% (v/v) ethanol. Crystals were cryoprotected by gradually supplementing the drop with
36 % (v/v) ethylene glycol (in 1 % steps, every 5 minutes) and flash frozen in liquid
nitrogen. Crystals of the CNC changed morphology after the inclusion of a second sAB
(sAB-87) and appeared in the orthorhombic space group P2
1
2
1
2
1
with one copy of the
"NB
-
NB
-87 complex in the asymmetric unit. Crystals grew to maximum
dimensions of ~50
×
100
×
150
μ
m
3
within 1 week.
Crystals used for diffraction
experiments were grown in 0.1 M MES, pH 6.5, 5 % (w/v) PEG 20000, and 20 mM
SrCl
2
. Crystal were cryoprotected by serial transfers into solutions containing 5 %, 10 %,
15 %, 20 % and 25 % (v/v) ethylene glycol supplemented reservoir solution.
This crystal
form typically diffracted to a resolution limit of ~9 Å. During a systematic heavy metal
derivative screen individual SeMet labeled crystals were identified that yielded X-ray
diffraction data to a resolution limit of 7.6 Å after soaking with 1 mM potassium
hexachloroosmate (K
2
OsCl
6
). Native crystals were derivatized by adding 0.2
μ
L of a
saturated tantalum bromide cluster (Ta
6
Br
14
) solution to the crystallization drop and
incubated for 1 week. X-ray diffraction data were collected at 100 K at beamline BL12
-2
at the Stanford Synchrotron Radiation Source (SSRL) and beamline GM/CA
-CAT 23ID-
5
D at the Advanced Photon Source (APS) on a Pilatus3 detector. Thousands of CNC
crystals were screened to yield the reported X-ray diffraction datasets.
Structure determination and model building.
X-ray diffraction data was processed
with XDS (
19
). The structures for both crystal forms were solved by iterative cycles of
molecular replacement (MR) using Phaser (
20
).
>ABCAD2CDA434C4A<8=0C8>=>5C7458ABC2AHBC0;5>A<2>=C08=8=6C74"NB
-57
complex in the space group C2, Phaser was run with the assumption that the asymmetric
unit (ASU) harbored one CNC
N
sAB-57 complex (~450 kDa), corresponding to a solvent
content of ~83 %. The crystal structures of the
S. cerevisiae
CNC components
'42 N"D?
NTD
$ #'47 N"D?
NTD
(PDB ID 3F3F), Nup120
NTD
(PDB
ID 3F7F), Nup84
NTD
(PDB ID 3IKO), and a structure of the sAB scaffold (PDB ID
3PGF) were used sequentially as search models with a model variance of 100 % sequence
identity (
6-10
). MR was performed in the above search order and the top solutions
were
taken from each MR search to look for the next molecule. During each MR round, Phaser
robustly obtained solutions with clear separation from other solutions after the packing
test with Log Likelihood Gain (LLG) values and refined translation function
Z-scores
(. >5 (. '42 N"D?
NTD
), (2) LLG=94, TFZ=11.6
'47 N"D?
NTD
), (3) LLG=190, TFZ=6.9 (Nup120
NTD
), (4) LLG=202, TFZ=8.1
(Nup84
NTD
), and (5) LLG=310, TFZ=8.5 (sAB scaffold) (fig. S1A, B).
The correctness of the final solution
output from Phaser was
assessed on the
following criteria: (1) clear separation of the best scoring solutions from the remaining
solutions at every step, (2) very high TFZ scores after each step, as TFZ scores above 8
usually indicate a definite solution,
(3) increasing LLG scores at each step indicating that
each additional molecule was improving the solution, (4) an internal test that the
Nup84
NTD
was placed in the same orientation as previously determined in the
'42 N"D? N"D?
NTD
crystal structure (
8
), despite no
a priori
information
restricting it to that location, (5) the overall shape of the solution was consistent with low
resolution EM reconstructions, and most importantly (6) the appearance of strong
additional features in the calculated electron density maps of the final solution. Strong
positive difference density for the helices of the triskelion were clearly visible in the
|
Fo
|
-
|
Fc
|
map output from PHENIX (fig. S2A). Density modification of the MR solution using
RESOLVE (
21
)
yielded an improved electron density map with additional density for
loops connecting the new helices despite no additional model building (
fig. S2A).
Furthermore, no additional density was observed in the solvent channels (
fig. S4A).
Model building was performed with COOT (
22
). The
α
-helical C-terminal domains of
Nup145C, Nup85, and Nup120 formed distinctive arrays of tubular electron density at
7.4 Å, into which we were able to place idealized
α
-helices. As the C-terminal domains
of Nup145C, Nup85, and Nup120 are connected to their respective N-terminal domains
by short loops, a preliminary model for the connectivity and directionality of the helices
was traced starting from the C-terminus of each previously determined structure. This
preliminary model was validated by comparison with the helical arrangement in the
S.
pombe
homolog of Nup120, which could be structurally aligned with the helices assigned
to the C-terminal domain of Nup120. Once all of the helices were successfully assigned
to each protein, the connectivity of all three proteins could be assigned with the aid of the
helix and loop lengths from a secondary structure prediction (
fig. S3D). As the electron
6
density does not possess features to assign the sequence register, the numbering in the
structure is approximate and only reflects the order and directionality of each helix and
thus we modeled the triskelion with the sidechains truncated at the C
β
position.
>ABCAD2CDA434C4A<8=0C8>=>5C74B42>=32AHBC0;5>A<2>=C08=8=6C74"NB
-
NB
-87 complex, which grew in the space group P2
1
2
1
2
1
, sequential Phaser searches
5>AC74'42 N"D?
NTD
N"D?
NTD
heterotrimer (PDB ID 3IKO), Nup120
NTD
(PDB
ID 3F7F), and the sAB scaffold structure (PDB ID 3PGF) produced clearly separated
solutions with the following scores: (1) LLG=239, TFZ=13.3
'42 N"D?
NTD
N"D?
NTD
), (2) LLG=653, TFZ=12.2 (Nup120
NTD
), and (3)
LLG=887, TFZ=11.2 (sAB scaffold, sAB-87) (fig. S1C, D). Despite exhaustive attempts
with both the normal Phaser pipeline and brute-force
translation and rotation searches, no
!& B>;DC8>=B F4A4 834=C85843 5>A '47 N"D?
NTD
(PDB ID 3F3F) and a second sAB
scaffold
(PDB ID 3PGF) for sAB-57
and these molecules are likely disordered in the
crystal. The resulting maps were comparable in quality to
those of the C2 crystal form
(fig. S2B). The arrangement of Sec13, Nup145C
NTD
, Nup84
NTD
, and the Nup120
NTD
in
C7458=0;!&B>;DC8>=8BC74B0<40BC70C8=C74BCAD2CDA4>5C74"NB
-57 complex
(fig. S6). The correctness of the solution was confirmed by the calculation of an
anomalous difference Fourier map using the phases from the MR solution, which
revealed peaks for 20 selenium sites and 1 Os site (fig. S3A-C). Improved phases were
obtained with MR-SAD in Phaser using phases from the MR solution and the 21
anomalous scatterers. Subsequent density modification revealed clear tubular density for
the triskelion helices, including density for the Nup85
CTD
, which could be readily docked
8= C74 B0<4 2>=5>A<0C8>=B 0B >1B4AE43 8= C74 "NB
-57 structure. Additional
confirmation of the correctness of the solution was obtained by calculating an anomalous
difference Fourier map using anomalous X-ray diffraction data obtained from a
"NB
-
NB
-87 complex crystal derivatized with Ta
6
Br
14
, which revealed 8
tantalum bromide cluster sites (fig. S3A-C).
Of the 20 selenium peaks observed, 11 aligned with the expected selenium sites in
the previously determined structures of Nup84
NTD
and
'42 N
Nup145C
NTD
. An additional
8 selenium peaks were present in the newly built helices of Nup85
CTD
and Nup145C
CTD
,
which were used to confirm the directionality and approximate sequence assignment of
the helices. The final selenium site aligned with the last methionine present in Nup85
NTD
,
but no additional sites were observed for th
4 A4<08=34A >5 C74 '47 N"D?
NTD
heterodimer (fig. S3A
">0338C8>=0;4;42CA>=34=B8CHF0BE8B81;45>A'47 N"D?
NTD
in
density modified maps either, despite room being available in the lattice to accommodate
the molecules (fig. S4B). Thus, this part of the structure is presumed to be disordered in
this crystal form.
Structure refinement.
Refinement of both structures was performed with heavy
restraints using PHENIX, with 1 group B-factor per residue with similarity restraints and
positional refinement with secondary structure restraints and reference model restraints
for the portions of the structure for which there were high-resolution structures (
23
). We
elected to use models re-refined by the PDB_REDO server, as they had superior
geometrical parameters to the previously deposited structures (
24
). The best strategy for
B-factor refinement was determined by comparing the results of test refinements using
the following strategies: 1 B-factor per residue with similarity restraints, 2 B
-factors per
7
residue
with similarity restraints, 1 B-factor per group, and 1 B-factor per group with
TLS parameters (fig. S5A, B). We additionally tested the output of a refinement strategy
of 1 B-factor per residue without similarity restraints to ensure that B
-factors were
meaningfully restrained. Refinement with 1 B
-factor per residue with similarity restraints
yielded the lowest R-factors and realistic B-factors that were smoothly distributed across
the model (fig. S5B). Therefore, we elected to use that strategy with no TLS parameters
for the final refinement. The final models of the C2 and P2
1
2
1
2
1
crystal forms yielded
average B factors for the overall model of 716.5 Å
2
and 536.5 Å
2
, respectively, with
comparable B factors for all protein chains (
fig. S5D, E). These B-factors include the
overall B of the crystal, as is the standard method of reporting B-factors in PHENIX. The
resolution limits for both data sets were determined by using the paired refinement
technique described by Karplus and Diederichs (
25
). Paired refinements were performed
in
0.2 Å steps
from 8.0 Å to 7.0 Å
and the resolution limits were selected conservatively
before resolution steps that did not improve the model (
fig. S5F, G). The final models of
the C2 and P2
1
2
1
2
1
crystal forms, refined to a 7.4-Å and
a 7.6-Å
resolution, yielded R
free
and R
work
values of 35.3 %, 33.0 %, and 34.7 %, 31.8 %,
respectively. The
stereochemical properties of the two structures were determined by MolProbity
(
26
). The
CNC complex structures reported here have similar Ramachandran statistics as the search
<>34;BDB435>A'42 N"
up145C
NTD
N"D?
NTD
'47 N
Nup85
NTD
, and Nup120
NTD
. The
newly built triskelion has perfect stereochemical parameters with no residues in the
disallowed region of the Ramachandran plot. For details of the data collection and
refinement statistics see table S2.
Analytical size-exclusion chromatography.
Protein-protein interaction experiments
were carried out on a Superdex 200 10/300 GL gel filtration column equilibrated in a
buffer containing 20 mM TRIS (pH 8.0), 100 mM NaCl, and 5 mM DTT. The various
combinations of the yeast CNC components were mixed and incubated for 30 minutes on
ice using a 1.2 molar excess of the smaller proteins. Complex formation was monitored
by injection of the pre-incubated proteins or the individual components onto the gel
filtration column. All proteins were analyzed under identical buffer conditions and
complex formation was confirmed by SDS-PAGE of the protein-containing fractions,
followed by Coomassie brilliant blue staining.
EM docking.
The crystal structure of the
yeast CNC was docked into the negative stain
EM reconstructions of the yeast complex (EMDB-5151) and the human complex
(EMDB-2443) using the Fit in Map function in the UCSF Chimera software package
(
27
).
The crystal structure of the yeast CNC was docked into the cryoelectron EM
tomographic reconstruction of the human NPC (EMDB
-2444) using an exhaustive,
unbiased six-dimensional search using a C
α
trace of the CNC structure with the program
ESSENS from the Uppsala Software Factory package RAVE (
28
). The rotations were
sampled in 10° steps across
α
,
β
, and
γ
for a total of 26,011 rotations, which were each
tested at all of the 366,980 grid points which had a map value greater than 1.5. Each
combination of rotation and grid point was scored by the K-minimum sum function over
the lowest scoring 60 % of the atoms against the average of
the 8 nearest grid points, as
implemented in ESSENS (
28
). This exhaustive scoring method produced a clear
8
separation of 65 top scoring placements from the remaining orientations (
fig. S12A). The
positioning of the top 65 placements in the EM reconstruction was further refined and
rescored using an orthogonal scoring method with the Fit in Map tool of UCSF Chimera
(
27
), which we used to calculate the cross-correlation of the EM map with a simulated
map calculated at 34 Å for each docked model (
fig. S12B).
Analysis of these solutions and their placements revealed a clear separation between
the top 32 solutions and the remainder of solutions.
Because of the eight-fold rotational
symmetry in the map, unique solutions are each composed of 8 solutions related by
rotational symmetry. As a result, the top 32 solutions form four unique rings, all of which
are
compatible when simultaneously placed into the NPC. The remaining solutions could
be classified as one of the following: (1) solutions that refined into one of the above
orientations upon refinement, (2) solutions with moderate scores lower than the top
scoring 32 orientations and could be discarded due to clashes (fig. S12C), or (3) low
scoring solutions that yielded much worse fits than the top 32 solutions upon refinement.
Despite the presence of additional features in the map for cytoplasmic filament
nucleoporins and associated mRNA export factors, the two CNC rings on the cytoplasmic
face and the two CNC rings on the nucleoplasmic face are identical (
fig. S12D). This
additional unbiased test was taken as final confirmation that this stoichiometry and
orientation of CNCs reflects their organization in the NPC.
Illustration and figures.
Structural figures were generated using PyMOL
(www.pymol.org).
9
Fig. S1.
Structure determination statistics of the yeast CNC.
(
A
)
Table of statistics for each
step of molecular replacement performed using Phaser for structure determination of the
CNC
N
sAB-57 complex. Each sequential step was performed with the top solutions from
the previous step, and a clear separation of the top solutions was apparent with each step.
(
B
) Plots of the initial log-likelihood gain (LLG) scores for the top 10 peaks from the
translation function step of Phaser for structure determination of the CNC
N
sAB-57
complex. (
C
)
Table of statistics for each step of molecular replacement performed using
10
Phaser for structure determination of the CNC
N
sAB-57
N
sAB-87 complex. Each sequential
step was performed with the top solutions from the previous step, and a clear separation
of the top solutions was apparent with each step. (
D
) Plots of the initial log-likelihood
gain (LLG) scores for the top peaks from the translation function step of Phaser for
structure determination of the CNC
N
sAB-57
N
sAB-87 complex. For each step, there were
only a handful of peaks selected from the rotation function, resulting in fewer peaks in
the translation function.
11
Fig. S2
Electron density during structure determination of the yeast CNC.
The electron
density for the initial molecular replacement and after density modification for the crystal
structures of the (
A
) CNC
N
sAB-57 and (
B
) CNC
N
sAB-57
N
sAB-87 complexes are shown.
Clear density was visible for the triskelion helices after successful placement of
previously solved crystal structures (left) and remained after density modification (right),
which was performed prior to model building.
The models visualized are the direct output
from molecular replacement prior to any interpretation.
12
Fig. S3.
/$"3*/./'"./-",/422$"33&1&12*.$1823",2/'3)& ;2
-
;2
-87 complex.
Anomalous difference Fourier maps were calculated for X-ray diffraction data collected
0CC74B4;4=8D<0=3C0=C0;D<?40:F0E4;4=6C7B5>A"NB
-
NB
-87 crystals grown
with SeMet labeled protein and soaked with K
2
OsCl
6
>A=0C8E4"NB
-
NB
-87
crystals soaked with tantalum bromide clusters (Ta
6
Br
14
). (
A
)
A ribbon representation of
C74 BCAD2CDA4 >5 C74 "NB
-
NB
-87 complex, with the anomalous difference
Fourier maps of
X-ray diffraction data collected at the selenium (purple) or tantalum
(blue) peak wavelengths contoured at 3.5
σ
and 4.5
σ
, respectively. Tantalum peaks
adjacent to Nup85
NTD
and a peak corresponding to the last selenium site in Nup85
NTD
are
visible despite the molecule being disordered in the crystal. (
B
) Close-up view of
selenium sites present for the triskelion helices for which there are no high-resolution
structures. 8 peaks are visible and confirm the positioning and orientation and
approximate sequence assignment in the structure. (
C
) Close-up view of the selenium
peaks in Nup84
NTD
and Nup145C, with stick representations of the SeMet residues
highlighting the expected sites. (
D
) Sequence and secondary structure prediction of
Nup145C
CTD
and Nup85
CTD
with methionine residues highlighted.
13
Fig. S4.
Crystal packing in crystals of
CNC
;2
-57 and
;2
-
;2
-87.
Representative views of the crystal packing for crystals of the CNC
N
sAB-57 complex (
A
)
and CNC
N
sAB-57
N
sAB-87 complex (
B
). (left) Uncarved density-modified electron
density contoured at 1
σ
demonstrates the large solvent channels present in both crystals.
(center) Ribbon representation of the asymmetric unit and surrounding symmetry mates
colored gray highlights a major crystal contact made in both crystals by a synthetic
antibody with the Nup145C
N
Nup84
NTD
interface. (right) Combined view of both the
electron density and unit cell.
14
Fig. S5.
Determination of optimal refinement strategy, resolution limits, and B-factor
analysis.
(
A
) Ribbon representations of the CNC
N
sAB-57 complex colored from blue to
red by B-factor for various alternative refinement strategies with histograms of the B-
factor distribution below. (
B
) Ribbon representation of the CNC
N
sAB-57 complex refined
with the final refinement strategy of 1 B-factor per residue and colored on the same scale
as in (A). (
C
) Ribbon representation of the CNC
N
sAB-57
N
sAB-87 complex refined with 1
B-factor per residue and colored on the same scale as in (A). (
D
-
E
) Average B-factors per
protein chain for the CNC
N
sAB-57 complex (
D
) or CNC
N
sAB-57
N
sAB-87 complex (
E
).
15
(
F
-
G
). Paired refinement analysis of the resolution limit as described by Karplus and
Diederichs (25)
for the (
F
) CNC
N
sAB-57 complex or (
G
) CNC
N
sAB-57
N
sAB-87
complex. The improvement in R-factors gained for each 0.2
Å
shell of data was assessed
by re-calculating the R-factors in the lower resolution
data after refinement with the
higher resolution data.