of 69
Title:
Architecture of the linker
-scaffold in the nuclear pore
Authors:
Stefan
Petrovic, Dipanjan
Samanta,
#
Thibaud
Perriches,
#
Christopher
J. Bley, Karsten
Thierbach,
Bonnie
Brown, Si
Nie, George
W.
Mobbs,
Taylor
A.
Stevens
, Xiaoyu
Liu,
André
Hoelz*
Affiliations:
California Institute of Technology, Division of Chemistry and Chemical Engineering, 1200 East California
Boulevard, Pasadena, CA, 91125, USA
#
these authors contributed equally to this work
*Correspondence: hoelz@caltech.edu
(A.H.)
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
Abstract
The nuclear pore complex (NPC) is the sole bidirectional gateway for
nucleocytoplasmic transport.
Despite recent progress in elucidating the arrangement of the
structured scaffold building blocks in the
NPC symmetric core, their cohesion by multivalent unstructured linker proteins
remained elusive.
Combining biochemical reconstitution, high resolution
structure determination, docking into cryo
-electron
tomographic reconstructions, and physiological validation, we elucidate
d
the architecture of the entire
linker
-scaffold, yielding a near-
atomic composi
te structure of the symmetric core accounting for ~77
MDa
of the human NPC.
Whereas linkers generally play a rigidifying role, the linker
-scaffold of the NPC
provides the plasticity and robustness necessary for the reversible constriction and dilation of i
ts central
transport channel.
Our results complete the structural characterization of the NPC symmetric core,
providing a rich foundation for future functional studies.
One sentence summary:
An
interdisciplinary analysis
established the near
-atomic
molecular architecture
and evolutionary
conservation of the linker
-scaffold of the human
nuclear pore complex.
2
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
Introduction
The
enclosure
of genetic material in the nucleus requires the selective transport of folded proteins and
ribonucleic acids across
the nuclear envelope, for which the nuclear pore complex (NPC) is the sole
gateway
(
1-4
).
Beyond its function as a selective, bidirectional channel for macromolecules, the role of
the NPC extends to genome organization, transcription regulation, mRNA maturation, and ribosome
assembly (
1, 2
).
The NPC and its components
are
implicated
in the
etiology of
many
human diseases,
including
viral infections
(
5, 6
).
The building blocks of the NPC are a set of ~34 different proteins
collectively termed
nucleoporins (nups). In the NPC, nups assemble into defined subcomplexes that are
generally present in multiples of eight, adding up to a mass of ~110
MDa in the human NPC
(
1-4
).
The
NPC architecture consists of a symmetric core with asymmetric decorations on its nuclear and
cytoplasmic faces
(Fig.
1A)
. The symmetric core displays
eight
-fold rotational symmetry
along
the
nucleocytoplasmic axis and two
-fold symmetry
along an axis
c
oplanar with
the nuclear envelope.
It
consists of two outer rings that sit on top of the nuclear envelope and an inner ring that lines the lumen
generated by the fusion of the two lipid bilayers of the nuclear envelope
. From the inner ring,
unstructured
phenylalanine
-glycine (FG) repeats are projected into
the central transport channel
to establish a diffusion
barrier that bars particles larger than
~40
kDa from freely diffusing across the nuclear
envelope.
Transport
is mediated by karyopherins, whose high
affinity for FG repeats and ultrafast exchange kinetics allow
karyopherin
-bound cargo to traverse the diffusion barrier
(
7- 10
).
The structural characterization of the NPC has progressed through
efforts to reconstitute and
crystallize ever larger portions of it, from small nup domain fragments to complexes as large as the
~400
kDa hetero
-heptameric Y-
shaped coat nup complex (CNC) (
11- 27
).
In parallel, progress has been
driven by efforts to push the resolution of cryo-
electron tomographic (cryo-
ET) reconstructions of intact
NPCs(
5, 6
)
. The docking of the CNC crystal structure into an ~32
Å cryo-
ET map of the intact human
NPC was the first demonstration that biochemical reconstitution and crystal structures could be used to
interpret cryo
-ET maps and unraveled the head
-to-tail tandem arrangement of CNCs in the outer rings.
The reconstitution and piecemeal structural analysis of
two hetero-
nonameric ~425 kDa inner ring
complexes
(IRCs)
provided the basis for docking 17 symmetric core nups into an ~23
Å cryo-
ET map of
3
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
the intact human NPC
(
25, 28
), yielding a near
-atomic composite structure of the entire ~56
MDa
symmetric core of the human NPC
(
29, 30
).
A similar approach elucidated the n
ear
-atomic architectures
of constricted and dilated states of the
S. cerevisiae
NPC
using nup and nup complex
crystal structures
to interpret ~25
Å cryo-
ET maps (
31, 32
).
Apart from an additional distal CNC rin
g and associated nups
present in the outer rings of the human NPC, the human and
S. cerevisiae
NPC
present equivalent nup
arrangements (
29- 33
).
The
inner ring of the human NPC
is com
posed of
six
scaffold nups
NUP155, NUP188, NUP205,
NUP54, NUP58,
and
NUP62,
two
primarily unstructured linker nups NUP53 and NUP98
, and NUP93,
which
is a hybrid of both (
29, 30
).
The doughnut
-shaped inner ring adopts a concentric cylinder
architecture,
in which
membrane
-anchored
NUP155 form
s
the
outermost
coat, followed by
layer
s
of
NUP93, NUP205/NUP188, and the NUP54
•N
UP58•N
UP62 channel nucleoporin
hetero-
trimer
(CNT)
in
the center, providing
the FG
repeats to form the diffusion barrier in the central transport channel
. Unlike
the
extensive
interactions of large
, folded domains
found in
the
CNC (
13- 17, 19, 22, 24, 26, 34
), the
structured do
mains of the
inner ring
nups do not
interact
directly.
Instead, the inner ring is
held
together
by
the
linker
nups
NUP53,
NUP98,
and an unstructured region of
NUP93, which
are proposed to
connect
the
scaffolds of the
four layers (
25, 28, 30, 35, 36
).
The resulting
linker
-scaffold
architecture allows for a
substantial ~200
Å dilation of the
inner ring’s
central transport channel
, accompanied by
the generation
of lateral gaps between the eight spokes, as observed in
recent
cryo
-ET analys
es of purified
and
in situ
human and yeast
NPCs (
31, 37-
40
).
The linker
-scaffold
is expected to play a
fundamental
role in
establishing an
architectural
framework to accommodate
the structural changes associated with the
reversible constriction and dilation of the inner ring.
Whereas our previous work achieved the
identification of the majority of the scaffold nup locations
in the NPC, a comprehensive understanding and the molecular details of the linker
-scaffold interaction
network that mediates the cohesion of the symmetric core has remained elusive. Here, we repo
rt the
characterization of all linker
-scaffold interactions through residue
-level biochemical mapping of
scaffold
-binding regions in linker nups and the determination of crystal and single particle cryo
-EM
structures of linker
-scaffold complexes.
Our analy
sis
revealed a common linker
-scaffold binding mode
,
4
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
whereby
linkers
are anchored by
central
structured mot
ifs
whose binding is reinforced by
disperse
interactions of
flanking regions.
We quantitatively docked the complete set of
linker
-scaffold
structures
into an
~12Å cryo
-ET reconstruction of the human NPC
(provided by Martin Beck)
(
41
),
and an
~25Å
in
situ
cryo
-ET
reconstruction of the
S. cerevisiae
NPC
(
31
).
In the inner ring,
our
new
linker
-scaffold
structures
allowed for
the
unambiguous assignment of
NUP188 and NUP205
to 16 peripheral and 16
equatorial positions, respectively.
From the nuclear envelope to the central transport channel, l
inkers
bridge
the
layers
of the inner ring to
coalesce scaffold nups into
eight
relatively rigid spoke
s that are
flexibly inter
-connected
, allowing for the formation of
lateral gates
. The
linker
-scaffold confers the
plasticity necessary for the reversible dilation and constriction of the inner ring’s central transport channel
in response to alterations in t
he nuclear envelope membrane tension.
The
topology of
linker
-scaffold
interactions between
inner ring
nups
is conserved from fungi to humans
. We carried out
systematic
functional analyses
of the linker
-scaffold network,
including the development of a
minimal linker
S.
cerevisiae
strain
, establishing its robustness and essential nature.
O
ur quantitative docking analysis
of the human NPC
revealed
eight NUP205
•NUP93
complexes
that
cross-
link
adjacent spokes in both
nuclear and
cytoplasmic outer rings
.
Facing the central transport channel,
an additional
eight
NUP205
•NUP93
copies
are exclusively anchored
at the base of the cytoplasmic outer ring
. NUP93
emerges
as a versatile
linker
-scaffold
hybrid
that recruits and positions the FG
repeat
-harboring
CNT to
the inner ring
and
reinforces th
e tandem
head-
to-tail
CNC arrangement
in the outer rings,
explaining
its
fundamental role in
mai
ntaining the
integrity of the entire NPC.
Our analysis completes the structural
characterization
of the
~77
MDa
symmetric
core and
lays out a
roadmap for future
studies on the NPC
assembly and
function.
5
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
RESULTS
Biochemical and structural analysis of
C. thermophilum
scaffold
-linker interactions
Nup192 and Nup188 are the scaffold keystones of two alternative eight
-protein inner ring complexes,
which both include the
linker
s Nup145N, Nup53
, and the scaffold
-linker hybrid Nic96
( Fig. 1,
B
to D
) (
26,
28, 30, 35, 36
). Previously, a composite structure of
the full
-length
~ 200
kDa
Nup192 was determined by
superposing
overlapping
structures of
its N
- and C
-terminal parts, revealing
an extended
a
-helical
solenoid
with a
question mark-
shaped architecture
, composed of five
Huntingtin,
elongation factor 3
(EF3), protein
phosphatase 2
A
(PP2A), and the yeast kinase
TOR1
(HEAT), fifteen
Armadillo (ARM)
repeats, and a
prominent
central
Tower
(
30, 35, 36, 42
).
A question mark shape was also observed for
Nup188 by negative
-stain electron microscopy (EM)
(
35
)
and high
-resolution
structures of its ~
130
kDa
N-
terminal domain (NTD), which contains a central SH3
-like domain insertion,
and
~45
kDa C-
terminal
Tail
region
again revealed extended
a
-helical solenoids composed of ARM and HEAT repeats
(
25, 43
).
However, no structural information could so far be obtained for the ~32
kDa Nup188 central region
equivalent to the Nup192 Towe
r.
Furthermore, although binding of Nup192 and Nup188 has
biochemically
been
mapped to
extensive
unstructured regions in t
he linkers
Nic96, Nup53 and Nup145N
,
the molecular details of these interactions have yet to be elucidated.
Nup192 interaction with Nic96
Having previously determined the crystal structure of a
C. thermophilum
Nup192 fragment only missing
the
N-
terminal Head subdomain (
Nup192
D
Head
, residues 153
-1756) (
30
),
we were able to
obtain
co-
crystals
with our previously mapped
Nic96
187-
301
fragment that
diffracted to 3.6
Å
resolution, allowing
for structure determination by
molecular replacement with the Nup192
D
Head
structure
as search model
(Fig. 2
). The Nic96 sequence register was unambiguously assigned by calculating
anomalous difference
Fourier map
s to locate seleno-L-
methionine (
SeMet)-l
abeled endogenous M263 and substituted A289M
residues (
fig.
S1). In the
Nup192
D
Head
•Nic96
187-
301
complex structure,
Nup192 adopts
the previously
described question mark shaped
a
-helical
solenoid
(
30
),
on which binding of
Nic96
(residues 240
-301;
R2)
extends from the
midpoint
to the base, burying ~3,700
Å
2
of combined surface area.
Nic96
R2
forms
6
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
two amphipathic
a
-helices
connected via a sharply kinking connector. T
he
longer N
-terminal
a
-helix is
cradled by an extensive concave Nup192 surface formed by ten ARM and HEAT repeats and the central
Tower, while the shorter
C-
terminal
one packs against
a hydrophobic patch
formed from the C
-terminal
Nup192
a
-helices
a
75-
77 (ARM20).
Nic96
residues 187
-239 were not resolved in the complex structure,
but were found to be dispensable for Nup192 binding by
Isothermal titration calorimetry (ITC),
as both
Nic96
187-
301
and Nic96
R2
retained
a dissociation constant (
K
D
) of
~75 n
M (
Fig. 2E, fig. S2
, A and D). To
obtain the structure of full
-length Nup192 with Nic96
R2
despite its resistance to crystallization, we
determined a single particle cryo-
EM reconstruction of the complex at 3.8
Å resolution from a refined set
of 176,609
particles (
Fig. 2B, fig.
S3
). The cryo
-EM reconstruction confirmed the molecular details of
Nic96
R2
binding, but showed a slightly wider gap
between the Nup192 Head subdomain and Tower
(fig. S4
). To validate the
molecular details of the Nup192
-Nic96
R2
interface
, we performed
structure
-guided
mutagenesis
and assessed
binding by
both
size-
exclusion chromatography coupled
to
multi
-angle light scattering (SEC
-MALS)
and ITC.
Consistent with
the extensive
hydrophobic nature of
the
interface
, binding was not strongly affected by
individual
substitutions, but abolished by triple
mutations in Nic96 (FFF mutant; F275E, F278E, and
F298E) (Fig.
2 , C,
E
and F,
figs. S5, S2B,
and
S6
)
or Nup192 (LAF mutant; L1584E, A1648E, and F1735E) (
Fig. 2
, D, E
and F
figs. S2C, S6,
and
S7).
Nup188 interaction with Nic96
Having identified the Nic96
R2
fragment as sufficient for Nup192 binding, we tested whether the same
Nic96 region was also sufficient for Nup188 binding (
Fig. 3
).
Indeed,
ITC measurements revealed that
Nup188 binds both Nic96
R2
and Nic96
187-
301
with similar
K
D
s
of ~90
nM (Fig.
3E
and fig.
S8, A
and D
).
We were able to determine a crystal structure of Nup188
•Nic96
R2
at 4.4
Å resolution as well as
a crystal
structure of
Nup188
NTD
(residues 1-
1134) at 2.8
Å resolution, which was used to aid with the phasing,
model building, and
sequence assignment. The Nup188 and Nic96
R2
sequence registers were
unambiguously assigned by identifying
SeMet
-labeled residues in
anomalous difference Fourier maps
(Fig. 3B, fig. S9
). Similar to the Nup192•
Nic96
R2
complex, Nup188 adopts an overall
question
mark
-shaped architecture, composed of an N
-terminal Head subdomain, 9 HEAT, 13 ARM repeats, and
7
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
a central, comparatively more compact Tower (
Fig. 3B), with
Nic96
R2
binding an ~3,700 Å
2
concave
surface
between the midpoint and the base of the Nup188 molecule. However, although Nup188
-bound
Nic96
R2
again forms two amphipathic
a
-helices, their secondary structure is switched compared to its
Nup192
-bound form (Fig.
3B)
: Whereas Nup192
-bound Nic96
R2
has a longer N
-terminal helix and a
shorter C
-terminal helix which wraps around the Nup192
Tail
, Nup188-bound Nic96
R2
has a shorter
N-
terminal helix that binds to the central Tower, and a longer C
-terminal helix cradled in the concave
surface at the base of
Nup188 (
Fig. 3B)
. Validation of the Nup188
•Ni
c96
R2
interface through
structure
-guided mutagenesis and binding assessment by SEC
-MALS and ITC again failed to show an
effect for single substitutions (Fig. 3
, Cand D
, figs.
S8,
S10,
and
S11). Remarkably, the Nic96
R2
FFF
mutation that abolishes
binding to Nup192 had the same effect on Nup188 binding, despite the
structural
polymorphism between
Nup192
- and Nup188
-bound
Nic96
R2
( Fig. 3
, C
to F, figs. S8, S
10
and
S12).
Analogous to the Nup192 LAF mutant, we were also able to identify a triple Nup188 mutation (FLV
mutant; F1478E, L1689E, and V1760E) that disrupted binding to Nic96
R2
( Fig. 3
, D
to F, figs.
S8
, S
11,
and S12).
Nup192 interaction with Nup145N and Nup53
To
identify the Nup145N
regions necessary and sufficient for
Nup192
binding, we used a combination of
five-
alanine scanning mutagenesis and truncation analysis
(
Fig. 4A)
. Substituting blocks of five
consecutive residues at a time to alanines, we found a hotsp
ot between residues 626
-655 that displayed
diminished affinity for binding to Nup192 when mutated (
Fig. 4A and fig.
S13). N-
and C-
terminal
Nup145N truncation resulted in a minimal Nup145N
R1
peptide (residues 616-
683) that recapitulated the
Nup192
-Nup145N
interaction (
Fig. 4B and fig.
S14).
Although shorter Nup145N fragments still bound to
Nup192 as well, they had progressively reduced affinity and ITC measurements confirmed that the
majority of the interaction is encapsulated in the Nup145N
R1
region, with
measured
K
D
s of ~825
nM and
~1,
500
nM for Nup145N and Nup145N
R1
, respectively (
Fig.
4G and fig. S
15).
Adding the newly mapped Nup145N
R1
and our previously mapped minimal Nup53
R1
fragment
(residues 31
-67)
(
36
)
, we reconstituted an ~
220
kDa Nup192
•Nic96
R2
•Nup145N
R1
•Nup53
R1
complex and
8
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
were able to obtain a single particle cryo
-EM reconstruction at 3.2
Å
resolution from a selected set of
484,910 particles (Fig. 4C
and
fig. S
16). The Nup145N
R1
binding site is
proximal to the Nic96
R2
binding
site, at the midpoint of the question mark
-shaped Nup192, where a hydrophobic pocket of
a-
helices
a
51,
a
53, and
a
54
anchors a Nup145N
R1
MYKL motif (residues 633
-636)
that runs perpendicular to the long
axis of the question mark, with its N
-terminus oriented towards the
N-
terminus
of Nic96
R2
( Fig. 4E). For
Nup53
R1
, the cryo
-EM map only resolved the central
phenylalanine
-glycine (FG) dipeptide, which bound
a hydrophobic Nup192 surface pocket at the top of the molecule formed by
a
-helices
a
19,
a
20, and
a
23.
Key contacts involve
L441
and
W499
of Nup192 and F48A of Nup53, consistent with our previous
observation that substitu
tion of either of these residues
abolishes the
interaction
s (
36
) (
Fig. 4D). Although
we previously established that basic regions around the central Nup53 FG
motif
are also required for
binding
(
36
),
the flanking residues were not resolved in the cryo
-EM density
for Nup53 or Nup145N.
Overall, comparison of the Nup192 structures in complex with different linkers demonstrates that linker
binding does not induce conformational rearrangements in the scaffold Nup192 (
fig. S
17).
Validation of the Nup192
-Nup145N
R1
interface through structure
-guided mutagenesis and
subsequent binding assessment confirmed the importance of the central hydrophobic
Nup145N
MYKL
motif (
Fig. 4F and fig.
S18), but complete ablation of binding was only observed when the three flanking
basi
c residues on either side were also mutated to alanine (10
-residue KKR-
MYKL-
RKR mutant; K624A,
K626A, R628A, M633A, Y634A, K635A, L636A, R643A, K650A, and R651A) (
Fig. 4, F
to H
, and
fig
s.
S15,
S18C,
and
S19). Reciprocally, targeting of the Nup145N
R1
MYKL binding site in Nup192
identified a quadruple mutant (LIFH; L1071A, I1075A, F1125A, and H1126A) that specifically abolished
Nup192 binding to Nup145N
R1
, but not Nic96
R2
and Nup53 (
Fig. 4
, F
and H
, and
fig
s. S
19
to S
21).
Nup188 interaction with Nup145N
To identify the minimal Nup145N fragment and key residues required for Nup188 binding, we refined our
previously identified binding region (residues 606
-750)
(
30
),
utilizing the same alanine scanning and
fragment truncatio
n approach used to map its binding to Nup192
. This
resulted in an extended
NUP145N
R2
region between residues 640 and 732, within which mutation of 706
-715 diminished
9
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
Nup188
NTD
binding (
Fig. 5
, A and B,
and
fig
s.
S22
and S23)
. Consistent with our
previous findings (
25
)
,
we also confirmed that Nup188 does not bind Nup53, even in the presence of Nic96
R2
and Nup145N
(fig.
S24
).
To structurally
characterize the Nup188
-Nup145N interaction, we
reconstituted a ~220
kDa
Nup188
•Nic96
R2
•Nup145N
R2
complex and determined its structure by single particle cryo
-EM.
An initial
set of 709,123 particles produced a reconstruction of Nup188•
Nic96
R2
at 2.4
Å
resolution that presented
weak excess density at the top of the question mark
-shaped Nup188 molecule (
Fig. 5C).
Further 3D local
classification identified a subset of 298,317 particles that yielded a reconstruction of
Nup188
•Nic96
R2
•Nup145N
R2
at 2.8
Å
resolution (Fig. 5C
and
fig.
S25). Nup145N
R2
buries residues I709,
L710 and F715 in a hydrophobic cradle formed by
a
-helices
a
22,
a
24, and the loop connecting
a
-helices
a
27-
a
28, braced by the SH3
-like domain. As with Nup192, only a portion of the Nup145N
R2
peptide was
resolved in the cryo
-EM density (residues 706
-718)
indicating that Nup188 also
forms multivalent
interactions
around a core Nup145N
R2
motif, enhanced by weaker binding events distributed across
flanking sequences on either side (
Fig. 5C).
As wi
th Nup192, no significant conformational
changes were
observed between
the different Nup188 structures
in response to linker binding
( fig.
S26).
Structure-
guided mutagenesis of Nup188 hydrophobic residues lining the
Nup145N
R2
-binding
cradle
identified a Nup188 HHMI mutant (H409A, H412A, M485A, and I535A) that abolished binding to
Nup145N
R2
, but not to Nup145N (Fig. 5Da and
fig.
S27
).
On the
Nup145N side, mutation of two
structurally resolved
residues,
L710A and F715A,
moderately disrupt
ed
Nup188
NTD
binding (
Fig. 5D
and
fig.
S28
).
Further systematic mutagenesis led to a Nup145N EDSILF mutant (E706A, D707A, S708A,
I709A, L710A, and F715A), which respectively abolished and reduced Nup188 binding to
Nup145N
R2
and Nup145N (
Fig. 5, E
and F,
and
figs.
S29
and
S30).
Overall, the greater tolerance of the Nup188
-
Nup145N interaction to binding site mutations compared to the Nup192
-Nup145N interaction
demonstrates an even greater reliance on promiscuous binding events in flanking regions
dispersed well
beyond the structurally resolved core.
10
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
Comparison of the Nup192-
and Nup188
-linker complexes
The determination of full
-length structures of both Nup192 and Nup188 scaffolds bound to their respective
linkers permits a direct comparison of these two distantly h
omologous
a
-helical solenoids (~28%
sequence similarity). Although both structures share the same overall question mark
-shaped architecture,
composed of HEAT and ARM repeats, the
Nup188
a
-helical solenoid displays
a tighter super-
helical
twist, resulting
in a
n ~10
Å
narrower
molecule with a
compacted upper ring (Fig.
6, A and B
). A Tower
protrudes from the midpoint of the
a
-helical solenoid towards the Head subdomain in both structures, but
extends further in Nup192 than the more compressed Nup188 version. Nic96
R2
binds both scaffolds at
the base of the question mark, but remarkably adopts different secondary structures to do so, switching
between which requires breaking and re
-forming
a
-helices (
Fig. 6C
).
By contrast, Nup145N binds to
different parts of the Nup192 and Nup188 question mark shape in the center or
at the top, respectively.
In contrast to Nup192, Nup188 does not bind Nup53, but, interestingly, the Nup145N
R2
binding site at the
top of Nup188 is nearly congruent with that of the Nup53
R1
binding site on Nup192
(Fig. 6A).
Our structures identify two
distinct types of linker
-scaffold interactions. Nic96
R2
binds with high
affinity,
utilizing the same well
-defined
~60-
residue motif
in binding to
both Nup192 and Nup188. On the
contrary, Nup145N binds to Nup192 and Nup188 through
extensive, overlapping,
binding regions and a
distinctive common binding mode, whereby both Nup145N interactions depend on a central
~10-
residue
anchor motif that is structurally well
-characterized, yet tight binding depends on extensive ~20
-60-
residue
N-
and C
-terminal flanking regions with high basic character.
A similar binding mode is also observed for
the Nup53
-Nup192 interaction. The observation that flanking regions
present in all three of these
linker
-scaffold interactions evade structural characterization suggests highly dynamic and promiscuous
interactions with the scaffold
surface. Notably, the previously characterized ~1
4- residue Nup170
-binding
motif of Nup145N (residues 729
-750) does not depend on binding
-enhancing flanking regions,
suggesting that the uncovered Nup145N/Nup53 mode of binding to the Nup192 and Nup188 scaffolds is
a desirable evolutionary outcome and architectural principle of the NPC inner ring.
Together, these data complete the structural characterization of the linker
-scaffold interactions
that
coordinate the larger structured nups
of
the inner ring into two
distinct
complexes, containing either
11
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
Nup192 or Nup
188.
Nup145N interacts at distinct sites on Nup192 and Nup188
and can only form
mutually exclusive interactions with either Nup192 and Nup170 or Nup188
, which
can be explained by
the significant overlap
in
the mapped binding regions for Nup192 (R1, 616
-683), Nup188 (R2, 640
-732)
and Nup170 (R3, 729
-750).
Whereas Nic96
R2
binds to large surfaces of both Nup188 and Nup192 with
~80
nM high affinity, binding of Nup145N and Nup53 utilizes
relatively short central
core
motifs that form
well
-defined interactions with scaffold surface patches and are enhanced by extensively
distributed
binding
through the
N-
and C
-terminal flanking regions.
This binding mode is reminiscent of Velcro, in
which the accumulation of weak
binding events builds towards a robust interaction that provides flexibility
because there are manifold productive binding configurations in terms of spatial distribution or occupancy
of binding residues. As an architectural principle, a Velcro
-like bindin
g between linkers and scaffolds is
advantageous in accommodating movements of scaffold domains without fully breaking linker
-scaffold
interactions, as the NPC expands and contracts.
Architecture of the
S. cerevisiae
linker-
scaffold
The inner ring of the NPC adopts an overall doughnut
-shaped architecture
with eight
-fold rotational
symmetry across a nucleocytoplasmic axis and two
-fold symmetry in the plane of the nuclear envelope
(
29, 30
).
Each of the 16 inner ring protomers were proposed to consist of a Nup192 inner ring complex
(IRC) and a Nup188 IRC, located
at the equatorial and peripheral positions, respectively
(
29, 30
).
Whereas quantitative docking of the folded scaffold nups
Nup170, Nic96, Nup192, Nup188 and CNT into
cryo
-ET maps of intact NPCs revealed their positioning to form four concentric cylinders, the linker
network that connects them has remained elusive.
Combined with our
previously determined
structures
of
Nup170•Nu
p53
R3
, Nup170•Nup145N
R3
, Nic96•Nup53
R2
, and Nic96
R1
•CNT
(
29, 30
),
the
Nup192•Nic96
R2
•Nup145N
R1
•Nup53
R1
and Nup188•Nic96
R2
•Nup145N
R2
structures now
allow us to
identify the locations of the entire linker
-scaffold network showing
all binding sites in the
NPC’s inner ring
(Fig. 1
, C
and D
).
With full
-length experimental structures available, we
statistically scored
the fit of
a million
randomly
placed
and
locally
refined
resolution-
matched
densities
simulated
from
12
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
Nup192
•Nic96
R2
•Nup145N
R1
•Nup53
R1
and Nup188
•Nic96
R2
•Nup145N
R2
structures against a
~
25
Å
in situ
cryo
-ET map of the
S. cerevisiae
NPC that captures a dilated inner ring state (
fig.
S31) (
31
).
Nup192 and Nup188 were
quantitatively
docked with high confidence
into inner ring equatorial and
peripheral question mark-
shaped cryo
-ET densities
, respectively, confirming previous proposals
(
31, 32
).
Structural differences, including the reduced width due to tighter superhelical winding of the Nup188
a
-helical solenoid and of the presence of an SH3
-like domain in Nup188, closely matched discriminating
features
of
the
cryo
-ET
density
(
fig.
S32).
Linker
-scaffold
Nup170
•Nup145N
R3
•Nup53
R3
,
Nic96
SOL
•Nup53
R2
, and CNT
•Nic96
R1
structures were placed by superposition with the previously
determined composite model of the
S.
cerevisiae
NPC inner ring
( Fig. 7
and fig.
S33) (
29, 30
).
The docking shows that Nup192 and Nup188 are linker
-scaffold hubs crucially positioned
between the membrane-
coating Nup157/Nup170 and the central transport channel
-interfacing CNT
layers of the inner ring.
The
placement of Nup170•
Nup145N
R3
•Nup53
R3
in the
Nup157/Nup170 layer
depicts it as a pegboard onto which linkers are anchored (
Fig. 7C)
. B
ecause the integrity of stochiometric
Nup192 and Nup188 inner ring
complexes
in solution
depends on linker
-scaffold interactions, we
wondered if the same interactions could be identified in the composite structure of the inner ring. Indeed,
Nic96
R2
bound to Nup192 and Nup188, connected to
respective equatorial and peripheral copies of
Nic96
SOL
and CNT
•Nic96
R1
( Fig. 7, E and F
). Nic96
SOL
-bound Nup53
R2
connected to Nup170
-bound
Nup53
R3
, Nup192
-bound Nup53
R1
connected to Nic96
SOL
-bound Nup53
R2
, and
Nup192
-bound
Nup145N
R1
connected to Nup170
-bound Nup145N
R3
( Fig. 7, D and E
). B
ecause linkers between the
different Nup145N scaffold binding regions in the same molecule are not sufficiently long to span the
physical distances between binding sites in
the composite structure, Nup188
-bound Nup145N
R2
and
peripheral Nup170
-bound Nup145N
R3
could not be connected to other inner ring Nup145N linker
-scaffold
binding sites, consistent with the mutually exclusive binding events observed in solution
(
30
).
Importantly,
the biochemical incompatibility corresponding to physically impossible connections suggests that the
assembly
of inner ring complexes
is
a priori
encoded by
their
biochemi
cal properties to prevent the
formation of
spurious linker-
scaffold complexes.
We considered whether
connections between linker regions bound to scaffold surfaces were
13
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
constrained to nearest neighbors by the physical length of the linker spanning two binding sites. We found
that to be true for the
peripheral and equatorial connections mediated by the N
-terminal Nic96 linker that
connects to Nup192/Nup188 and the CNT
(fig.
S34) and the Nup145N connection between Nup192 and
the equatorial Nup170 (
fig.
S34F
). As explained above, Nup188 and the
peripheral Nup170 each are
required to bind
their own Nup
145N copy
. Although a stretched Nup53 linker
can
reach
at least three
different
Nup170
positions
from each peripheral and equatorial Nic96
SOL
, the nearest neighbor is only
~50 Å
away in both cases, in contrast to all the other copies that are at least twice as distant.
Therefore,
Nup53 is expected to connect the peripheral Nup170 with the equatorial Nic96
SOL
, and
vice versa
,
the
equatorial Nup170 with the peripheral Nic96
SOL
, both across the
inner ring midplane (
fig.
S34, B and C
).
The
Nup53
-mediated
connection
between
Nup192
and
Nic96
SOL
is the only
linker
-scaffold interaction that
occurs between adjacent spokes (
fig.
S34G
), as
we
previously proposed
(
30
).
This linker
-scaffold
arrangement suggested
the Nup53
RRM
homodimer
is placed between spokes
, despite the lack of
corresponding cryo
-ET density
.
Together, these data elucidate the architecture of the
S. cerevisiae
inner ring linker
-scaffold. Apart
from Nup53-
mediated spoke bridging, all linker
-scaffold connections in the
S. cerevisiae
inner ring occur
within the same spoke, thereby allowing inter
-spoke gaps to form
. Thus, the linker
-scaffold
architecture
provides a molecular explanation for the inner ring’s ability to exist in
constricted
and dilated states
(
31
).
Future work needs to address the forces that govern the inner ring’s reversible dilation and
constriction
,
with
membrane tension
being
the most likely candidate
.
S. cerevisiae
linker-
scaffold is robust and essential
Due to
ancestral gene duplication events in
S. cerevisiae
, there are several linker and scaffold nup
paralogs,
including Nup145N paralogs Nup116 and Nup100, Nup53 paralog Nup59, and Nup170 paralog
Nup157
(
2
). Evolutionary retention of distinct inner ring paralogs suggests
they are either speciated to
occupy unique positions and roles within the assembled NPC, or are broadly redundant and capable of
accommodating several binding functions at once. A multispecies sequence alignment uncovered that
the
S. cerevisiae
Nup100 and
Nup116 paralogs contained
sequences homologous to the
Nup192,
14
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
Nup
188,
and Nup170 binding
regions characterized in
C. thermophilum
Nup145N
, but only Nup116
possesses
the Gle2 binding site (GLEBS) motif that recruits the mRNA export factor Gle2 (
fig.
S35) (
44
).
In addition, all three paralogs contain C
-terminal domains (CTDs) that were previously shown to form
interactions with either Nup145C or Nup82, which are components of the CNC and cytoplasmic filaments,
respectively
(
30, 45, 46
).
To interrogate the function of individual
scaffold
binding regions
in the linker
Nup116, we
first
established a
S. cerevisiae
minimal
linker strain
in which mutations would not be complemented by
the
Nup145N and Nup100
paralogs (
Fig.
8). Because Nup145N results from the cleavage of a Nup145
precursor prote
in whose other product Nup145C incorporates into the CNC and is required for survival
(
47
),
all experiments involving Nup145 deletion were carried out in the presence of
NUP145C
ectopically
expressed from a centromeric plasmid.
Building on previous studies of synthetic lethality
(
48
),
we
systematically established that all
double knockout combinations of
NUP100
,
NUP116
and
NUP145
can
be rescued by a chimeric
NUP116
-NUP145C
( fig. S
36)
.
Consequently, we found that
NUP116
is able to
complement the
nup100
D
nup116
D
nup145
D
strain
(in the presence of ectopic
NUP145C
), albeit with
background nuclear mRNA
retention above
wildtype
levels
( Fig. 8E
, fig
s. S36, C and D,
and S
37).
Unlike
previously described for Nup100
(
44
),
Nup145N c
himeras containing either a GLEBS motif or the entire
GLEBS-
containing
FG
-repeat
domain
of
Nup116
could
not
rescue
the
nup100
D
nup116
D
nup145
D
/NUP
145C
strain, suggesting an essential function for the R1/R2 regions that
are uniquely absent in Nup145N but not
the other paralogs (
fig. S
36, E
and F
).
Based on this result, w
e systematically mutated all functional elements in the Nup116 sequence,
and found that the GLEBS, FG and CTD are essential (
Fig.
8, A and B,
and fig.
S37, A
to
C). We targeted
the
Nup192, Nup188 and Nup170
scaffold binding regions (R1, R2 and R3, respectiv
ely) with three types
of mutations: deleting entire binding regions (
D
R1
,
D
R2
, and
D
R3
), replacing entire binding regions with
a glycine-
serine (GS)-linker of equivalent length (
R1/40×GS
,
R2/40×GS
, and
R3/12×GS
), or substituting
sequence-
conserved residues shown to disrupt binding
of the C.
thermophilum
Nup145N to the
respective scaffolds
(
R1m
,
R2m
, and
R3m
) (Fig. 8A
and
fig.
S35). All Nup116 variants were expressed
at comparable levels (
fig. S
37, C and E
).
After establishing viability, non
-lethal mutants were further
15
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint
assayed for growth rates at different temperatures, localization of eGFP
-Nup116, and
nuclear
retention
of mRNA
and 60S pre
-ribosomes. Mutations in each of the Nup192, Nup188 and Nup170 binding r
egions
resulted in altered phenotypes, albeit to different extents
(Fig. 8
and
fig. S
37). Deletions and GS-
linker
replacements, being aggressive types of mutations, were lethal if targeting R1, and impacted growth and
mRNA/60S pre
-ribosome export if introduce
d in R2 and R3. The less aggressive combination of
substitutions,
R1m
, allowed us to perturb R1 with significant yet non
-lethal phenotypic effects, which were
further exacerbated through combination with
R2m
(
R1m+R2m
) or R3m (
R1m+R3m
), culminating with
the
lethal
R1m+R2m+R3m
triple mutation
(Fig. 8
and
fig. S
37). Interestingly, all Nup116 mutations
resulted in temperature
-dependent loss of eGFP-
Nup116 from the nuclear envelope rim and concomitant
emergence of eGFP
-Nup116 foci, as previously reported (
Fig. 8C
and
fig. S
37F
) (
45, 49, 50
).
These
results
demonstrate the essential linker function of Nup116 in the
S. cerevisiae
NPC and
illustrate
the
physiological relevance of the biochemically and structurally characterized Nup192, Nup188 and Nup170
binding regions
.
Perturbing the linker
-scaffold interactions on the scaffold side, absence of Nup192 results in
lethality, matching
the severe consequences of deleting the Nup192 binding region in Nup116 (
Fig. 9B
).
Additionally, we had previously found that truncation of just the Nup192 Tail region has a deleterious
effect, attributable to the perturbation of the Nic96
R2
binding site
(
36
).
Our new Nup192
-Nic96
R2
structures
provide a molecular explanation for this result, showing Nic96 binding at the Nup192 Tail and the base
of the Tower (Fig.
6, A and B).
Based on sequence conservation and the insight from our structural and
biochemical characterization, we transposed the
C. thermophilum
Nup192
LAF
and
LIFH
substitutions
that ablated binding to Nic96
R2
and Nup145N
R1
, respectively,
in vitro
into
S. cerevisiae
Nup192, along
with designing aggressive
D
Tail -Tower
and
Tail -Tower
truncations deleting the respective linker binding
subdomains (Fig. 9
A
and fig. S
38). Individually, the
LAF
and
LIFH
mutations had no effect on growth,
although wildtype Nup192 could outcompete both for localization to the nuclear envelope rim. However,
the combination of
LAF
and
LIFH
substitutions (
LAF+LIFH
) and both Nup192 truncations failed to rescue
the lethal
nup192
D
phenotype (
Fig. 9
, A
to D, and fig. S
39).
Because
NUP188
is not an essential gene, we analyzed the effect of transposed Nup188
FLV
16
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted October 26, 2021.
;
https://doi.org/10.1101/2021.10.26.465796
doi:
bioRxiv preprint