Supplementary Information for
Structural and functional analysis of mRNA export regulation by the nuclear
pore complex
Daniel H. Lin, Ana R. Correia,
1
Sarah W. Cai,
1
Ferdinand M. Huber,
Claudia A. Jette, André Hoelz*
Affiliation:
California Institute of
Technology, Division of Chemistry and Chemical Engineering, 1200 East California
Boulevard, Pasadena, CA, 91125, USA
1
These authors contributed equally
*Correspondence
: hoelz@caltech.edu (A.H.)
Supplementary Notes 1
-7
Supplementary Figures 1
-21
Supplementary Tables 1
-6
Supplementary References
1
Supplementary Note 1:
Identification of a minimal Nup42 fragment sufficient for Gle1 binding.
We utilized
a haploid
nup42
∆
S. cerevisiae
strain harboring a genomic C
-terminal Gle1
-GFP fusion
analogous to a previously
reported strain.
This
nup42
∆
/gle1
-GFP
strain displays a
temperature
-sensitive phenotype causing
mislocalization of Gle1 from the nuclear rim under heat shock conditions
1
. W
e monitored the localization of Gle1
-
GFP and mCherry
-HA
-tagged Nup42 variants: full
-length Nup42 (residues 1
-4 30), Nup42 lacking the entire N-
terminal FG repeat region (residues 364
-430), two further N
-terminally truncated variants (residues 397
-430 and
410-
430), and Nup42 containing the FG repeats but harboring a 36
-residue C
-terminal truncation (residues 1
-
394).
Consistent with previous reports, these strains were viable, and the deletion of Nup42 did not affect Gle1
localization at 30
°C (
Fig. 2a; Supplementary Fig. 3a,b
)
1
. Nup42 truncations that contained
residues 397
-430
displayed nuclear rim staining consistent with localization to the NPC, whereas a Nup42
variant containing only
residues 410
-430 did not (
Fig.
2a).
Strains harboring m
islocalized Nup42 variants also displayed a temperature
-
sensitive growth phenotype at 37
°C along with a loss of Gle1
-GFP from the nuclear rim following a 42
°C heat
shock (
Fig.
2a; Supplementary Fig. 3b).
Furthermore, these data suggested that the Nup42
GBM
-Gle1 interaction
was not only required for Nup42 localization to the NPC, but also for maintenance of Gle1 localization at the
NPC during heat shock.
We note that the fluoresc
ent protein fusions contributed to the growth phenotype; the
Nup42
-dependent phenotype at 37
°C
was only observed for strains harboring a genomic C
-terminal Gle1
-GFP
fusion, in agreement with previous reports that Nup42 is not required in wild
-type yeast at 37
°C
2
.
Moreover,
given
the artificial nature of the
nup42
∆
/gle1-
GFP
strain and the primarily unstructured Nup42 N
-terminal region,
we cannot exclude that other regions on Nup42 may contain additional binding sites for other nucleoporins.
Supplementary Note 2: Details
of interactions between Nup42 and Gle1.
The recognition of Gle1 by Nup42
is primarily mediated by a c
onserved core set of hydrophobic interactions supplemented by additional peripheral
interactions.
Specifically, the
S. cerevisiae
Nup42 hydrophobic core contains two proline residues (P420 and
P424) that point their pyrrolidine rings inwards towards phenyl
alanine residues F409 and F414 (
Fig. 2h). This
core wraps around and buries the exposed hydrophobic Gle1 residues W451, Y488, and L495 (Fig.
2h). This
hydrophobic interface is supplemented by polar interactions, including a salt bridge between Nup42
residue
D421 and Gle1 residue R456, as well as a network of hydrogen bonds between Gle1 residues Q491 and K494
2
with the backbone carbonyls of Nup42 residues I408, F409, A411, and L428 (
Fig. 2h). Furthermore, these
interactions are highly conserved:
human G
le1 residues W602, Y637, M644 are recognized by human Nup42
residues F401, F406, P412, and P416;
C. thermophilum
Gle1 residues W447, Y484, and A491 are recognized
by
C. thermophilum
Nup42 residues W530, F539, P544,
and P548.
C. thermophilum
Nup42
GBM
is ext
ended by
24 additional N
-terminal residues which increase the interaction surface area (
Fig. 2f; Supplementary Fig. 11a).
This insertion appeared to be present at the sequence level among species within the
Pezizomycotina
subphylum of fungi (
Supplementary
Fig. 9
).
Supplementary Note 3:
C. thermophilum
IP
6
-binding pocket
.
To
directly assess the conformational changes
in Gle1
CTD
induced by IP
6
binding independent of Dbp5 binding, we
compared our
apo
and IP
6
-bound
C. thermophilum
Gle1
CTD
•Nup42
GBM
structures
( Supplementary Fig. 7a
). Minimal conformational changes
occur
upon IP
6
binding
, mostly limited to a loop directly adjacent to the IP
6
pocket (
Supplementary Fig. 7a
). In the
C. thermophilum
Gle1
CTD
•Nup42
GBM
•IP
6
s truc
tu re, two IP
6
molecules coordinating two Z
n
2+
ions from the
crystallization buffer bind simultaneously to each Gle1
CTD
. Multiple IP
6
molecules were also observed in
structures of
S. cerevisiae
Gle1
CTD
determined in the presence of IP
6
(without divalent cations) and the IP
6
primary binding site overlaps directly for the two species (
Supplementary Fig. 7c
)
3
. Thus, d
espite differences in
the location of positively charged residues in th
e
S. cerevisiae
and
C. thermophilum
IP
6
pockets,
the
electrostatic
potential of the pocket
was
conserved
and IP
6
bound in a similar orientation
, suggesting that IP
6
could function
in a similar role
in
C. thermophilum
as in
S. cerevisiae
( Fig. 3a,b,d,e;
Supplementary Fig. 7d
).
Supplementary Note 4:
Differences between human and yeast DDX19/Dbp5 complexes
. Our sequence
conservation analysis indicated that the IP
6
-binding lysine residues in the C
-terminal
a
-helix of
S. cerevisiae
Dbp5 are
not conserved in human DDX19 (
Supplementary Fig. 12b
). Instead, the DDX19 C
-terminal
a
-helix is
two helical turns shorter and packs more closely to Gle1
CTD
to position D470, D472, and E475 to form salt bridges
in the second interface (
Fig. 5f
). Another difference in human
Gle1
CTD
is that helix
a
2 curves to allow K416 and
K419
to form salt bridges
with DDX19
CTD
and
other residues to form more contacts with DDX19
CTD
. Additionally,
a 12-
residue
b
-tongue insertion between helices
a
3 and
a
4, unique to human Gle1
CTD
, also form
s new
contacts
3
with DDX19 (
Fig. 5f
). Altogether, the human proteins utilize a more extensive interface to compensate for the
absence of IP
6
binding.
Supplementary Note 5: Conformational changes in DDX19 induced by Gle1 binding
.
First,
the C
-terminal
helix became ordered, positioning the acidic residues of the helix to interact with lysines on the Gle1
CTD
surface,
resulting in the formation of interface 2 (
Fig. 5c,6a; Supplementary Movie
2
). In our structure of
apo
DDX19
∆
N53
(AMP
-PNP•Mg
2+
), the C
-terminal hel
ix is already partially ordered, likely because of crystal contacts
(Supplementary
Fig.
16). The second difference was a conformational rearrangement in the adjacent “trigger
loop” (residues 328
-335), allowing I331 to pack against the C
-terminal helix, T33
2 to pack against Gle1, and
moving Q335 out of the way of the neighboring “anchor loop” (residues 390
-403) (
Fig. 6b; Supplementary Movie
2). This rearrangement was reinforced by the extensive hydrogen bond network formed with Gle1
CTD
residues,
resulting in
the formation of interface 1 (
Figs. 5b,6b; Supplementary Movie 2
). Third, there was a large
rearrangement of the DDX19 anchor loop that led to a shift in register (see residue C393) and movement away
from the auto
-inhibitory helix (
Fig. 6c; Supplementary
Movie 2
). This loop rearrangement removes several
contacts between the auto
-inhibitory helix and DDX19
CTD
, including a salt bridge between D398 and R67 and
several hydrophobic interactions (
Fig. 6c
), which provides an explanation for the partial separation
of DDX19
NTD
and DDX19
CTD
. Lastly, the loop containing DEAD
-box motif VI (residues 429-
435), which was disordered in both
the
apo
DDX19
∆
N53
(ADP) and DDX19
∆
N53
(AMP
-PNP•Mg
2+
) structures, became ordered. Motif VI contains several
residues that directly bind A
TP in the closed, active conformation (R429, R432, and F433), and they adopt very
similar conformations to those observed in the structures of DDX19
∆
N53
(AMP
-PNP•Mg
2+
)•RNA (
Fig. 6d;
Supplementary Movie 2). Importantly, the Gle1
CTD
residues involved in the D
DX19
CTD
-binding
interfaces
are also
involved in stabilizing
conformational changes in DDX19
CTD
through hydrogen bonds and salt bridges, explaining
the sensitivity of DDX19 activation to the single point mutations we made in Gle1
CTD
( Fig. 5e). In summary,
Gle1
CTD
binding to DDX19
CTD
causes a cascade of conformational changes that partially releases the auto
-
inhibitory helix and prepares the residues involved in nucleotide binding to form the closed, active conformation.
4
Supplementary Note 6: DDX19 variants.
To obtain a better understanding of the role of various regulatory
elements in the N
-terminal region of human DDX19, we tested the activation of a series of truncation mutants
that included DDX19
∆
N53
, the crystallized construct, which still contained the
auto-
inhibitory helix; DDX19
∆
N67
,
which additionally removed the auto
-inhibitory
helix; and DDX19
∆
N91
, which further
removed mobile residues that
were a part of DDX19
NTD
and was analogous to the previously used yeast Dbp5 truncation construct. Using our
human DDX19
crystal structures, we also designed a mutant variant of full
-length DDX19, DDX19
S60D/K64D
,
containing two aspartate substitutions in the auto-
inhibitory helix
that we predicted would disfavor formation of
the inhibited state due to electrostatic repulsion.
Supplementary Note 7:
How can Nup214
NTD
binding be both inhibitory and stimulatory?
We propose that
Nup214
NTD
binding also favors the separation of DDX19
NTD
and DDX19
CTD
due to steric clashes between
Nup214
NTD
loops and DDX19
CTD
in the closed DDX19 conformations
4,5
. In previous structural work, we
crystallized Nup214
NTD
with full
-length DDX19, but
found
only Nup214
NTD
and DDX19
NTD
to be ordered in the
crystal structure due to the separation of the auto
-inhibitory helix and DDX19
CTD
from DDX19
NTD
in that crystal
4
.
Nup214
NTD
also inhibited the activity of the hyperactive DDX19 mutants even in the absence of RNA
(Supplementa
ry Fig.
18), suggesting
that Nup214
NTD
binding to DDX19
NTD
also prevents the formation of the
closed, catalytically-
competent conformation of DDX19. A similar role in separating yeast Dbp5
NTD
and Dbp5
CTD
was previously proposed for yeast Nup214
NTD
, based o
n the crystal structure of yeast
Gle1
CTD
•IP
6
•Dbp5
∆
N90
(ADP)•Nup214
NTD
, in which the Dbp5
NTD
is rotated further away from Dbp5
CTD
than in the
structure of yeast Gle1
CTD
•IP
6
•Dbp5
∆
N90
(ADP)
3
. Depending on which step is rate limiting, the separation of
DEAD-
box helicase domains may slow or enhance the DDX19 reaction rate.
Notably, Gle1
CTD
binding also inhibited the
hyperactive DDX19 mutan
ts, indicating that Gle1 binding also has multiple consequences on the DDX19 activity
cycle (
Fig. 7a). Further studies will be necessary to elucidate the precise molecular order of events in the DDX19
reaction cycle.
5
6
Supplementary Figure
1: The
interactions of Nup155
CTD
with Nup98
∆
FG
or Gle1
N
are mutually exclusive.
(a) Size exclusion chromatography analysis (SEC) of the interactions between Nup155
CTD
•
SUMO
-Gle1
N
and
Nup98
∆
FG
as
in Figure 1c,
but
with addi
tional control elution profiles. Purified
Nup155
•
SUMO
-Gle1
N
complex was
mixed with the indicated amounts of Nup98
∆
FG
and loaded on a Superdex 200 10/300
GL
column.
The
gray
horizontal bar
indicate
s the fractions visualized with Coomassie
-stained SDS
-PAGE gels.
(b
) SEC
of the
interactions between N
up155
CTD
•
Nup98
∆
FG
and SUMO
-Gle1
N
. Purified Nup155
•
Nup98
∆
FG
complex was mixed
with the indicated amounts of SUMO
-Gle1
N
and loaded on a Superdex 200 10/300
GL
column.
Control elution
profiles of
Nup155
•
SUMO
-Gle1
N
, Nup98
∆
FG
and
SUMO
-Gle1
N
are included for reference.
(c ) Purified
Nup155
•
SUMO
-Gle1
N
complex was mixed with
2- fold molar excess Nup214
NTD
and loaded on a Superdex 200
10/300
GL
column.
No displacement is observed, demonstrating the specificity of the mutual
exclusivity.
(d) GST
pull-
down experiments with GST-
Nup155
CTD
. Complexes of
(left)
GST
-Nup155
CTD
•
Nup98
∆
FG
or (right)
Nup155
•
SUMO
-Gle1
N
were
preformed
with a molar ratio of 1:2 and incubated for 30 minutes. After preincubation,
the preformed complexes were
incubated
for 30 minutes
with increasing concentrations of
(left)
SUMO
-Gle1
N
or
(right
) Nup98
∆
FG
( molar ratios of 0.5, 1, 1.5, 2, 4, and 8)
. Samples were then pulled down with
25
μ
l of glutathione
-
coupled sepharose beads
. Top gels contain the
40 % of the loaded sample, middle
gels contain 40 % of the
sample that was not pulled down, and
bottom gels contain the glutathione
-bound fractions.
7
8
Supplementary Figure
2: Identification of residues involved in the interactions
of
Nup98
∆
FG
and
Gle1
N
with
Nup155 binding.
(a, b)
SEC
analysis of Nup155
CTD
mutants for Nup98
∆
FG
and
SUMO
-Gle1
N
binding,
respectively.
Purified Nup155
CTD
mutants were
preincubated with (a)
Nup98
∆
FG
or (b)
SUMO
-Gle1
N
and loaded
on a Superdex 200 10/300 GL column.
Control SEC profiles
are shown for
Nup155
CTD
(black), SUMO
-Gle1
N
or
Nup98
∆
FG
( gray
) and
wild
-type complex
(blue). SEC profiles of Nup155
CTD
mutants preincubated with either
SUMO
-Gle1
N
or Nup98
∆
FG
are colored green for wild
-type levels of complex formation or orange for reduced
binding
. The
gray h
orizontal bar
s in chromatograms indicate
the fractions visualized with Coomassie-
stained
SDS-
PAGE gels.
(c)
SEC
analysis of the interactions between SUMO
-Gle1
N
alanine mutants and Nup155
CTD
.
Purified Nup155
CTD
was mixed with the indicated SUMO
-Gle1
N
mutants and loaded on a Superdex 200 10/300
GL column. SEC profiles of Nup155
CTD
(black), SUMO
-Gle1
N
(gra
y) and Nup155
CTD
preincubated with SUMO
-
Gle1
N
(blue) are shown as controls. SEC profiles of SUMO
-Gle1
N
alanine mutants preincubated with Nup155
CTD
are
colored
green for wild
-type levels of complex formation, orange for reduced binding, or red for
complete
disruption.
Asterisks indicate degradation products of SUMO
-Gle1
N
variants.
9
10
Supplementary Figure
3: Analysis of interaction between Nup42
GBM
and Gle
1
CTD
in
S
.
cerevisiae
.
(a)
Growth analysis of
S. cerevisiae
nup42
Δ
/gle1-
GFP
strains containing
the indicated
Nup42
-mCherry
-HA
variants.
10-
fold serial dilutions
were spotted onto SDC-
LEU plates and grown for 4 days at 30
°C
and 37
°C.
Schematics
on
the right indicate the
domain boundaries
of the Nup42 variants.
A solid line indicates omitted
regions
. (b)
I n
vivo
localization analysis in
S. cerevisiae
of Gle1-
GFP and Nup42
-mCherry
-HA
variants. C
onstructs are the
same as
in panel a
. Scale bar is 5
μ
m.
(c)
Western blot analysis of the expression levels of Nup42
-mCherry
-HA
variants in a
S. cerevisiae
nup42
Δ
/gle1-
GFP
strain
. Nup42
-mCherry
-HA
variants
and the hexokinase loading
control were detected with anti
-HA and anti
-hexokinase antibodies, respect
ively. Asterisks (*) indicate
nonspecific band
s detected by the antibodies.
11
12
Supplementary Figure
4: Pelleting t
hermostability assay.
(a)
S. cerevisiae
Gle1
CTD
or
Gle1
CTD
•
Nup42
GBM
was incubated at the indicated temperatures in the absence or presence of I
P
6
for 30 minutes prior to
centrifugation
. (b
)
H. sapie
ns
Gle1
CTD
or
Gle1
CTD
•
Nup42
GBM
was incubated at the indicated temperatures in the
absence or presence of IP
6
for 30 minutes prior to centrifugation. Pelleted (P) and soluble (S) fractions were
analyzed by SDS-
PAGE and visualized by Coomassie staining. Red arrows indicate the temperature at which
more than 50
% of total Gle1
CTD
pelleted
.
13
14
Supplementary Figure 5
: S
urface properties of
S. cerevisiae
Gle1
CTD
.
Surface representations of Gle1
CTD
in
four orientations related by 90° rotations. The Nup42
GBM
, IP
6
, and Dbp5 binding interfaces are outlined in black.
(a ) Identification of
S. cerevisiae
Gle1
CTD
binding surfaces.
The
IP
6
binding site is colored in yellow, Dbp5 binding
interface is colored in purple
, and Nup42
GBM
interface is colored in orange. (b
) Surface representation colored
according to sequence conservation for
fungi belonging to the
Sacchar
omycotina and Schizosaccharomycetes
groups using an alignment containing the species
S. cerevisiae
,
Z. rouxii
,
K. lactis
,
C. albicans
,
Y. liplytica,
T. deformans, S.
complicata,
and
S. pombe
. (c
) Surface representation colored according to electrostatic
potential from
-10 k
B
T/e (red)
to 0 k
B
T/e (white)
to +10
k
B
T/e (blue).
15
16
Supplementary Figure 6
: Analysis of interaction between
scNup42
GBM
and scGle1
CTD
.
(a,b)
SEC
analysis
of the effect of mutations in
scNup42
GBM
on
scGle1
CTD
binding.
Purified
scGle1
CTD
was
mixed with the indicated
scNup42
GBM
mutants and loaded on a Superdex
75
10/300 GL column. SEC profiles of
scGle1
CTD
(black),
scNup42
GBM
( gray
), and
scGle1
CTD
preincubated with
scNup42
GBM
(blue) are shown as controls. SEC profiles of
scNup42
GBM
mutants preincubated with
scGle1
CTD
are colored green for wild
-type levels of complex formation,
orange for reduced binding, or red for
complete disruption. The
gray
horizontal bar indicates the fractions
visualized with Coomassie
-stained SDS
-PAGE gels.
17
18
Supplementary Figure 7
: Analysis of the effect of IP
6
binding on Gle1
CTD
in fungi.
(a ) Comparison of the
C.
thermophilum
Gle1
CTD
•Nup42
GBM
•IP
6
and Gle1
CTD
•Nup42
GBM
structures. (b)
Comparison of the
S.
cerevisiae
Gle1
CTD
•Nup42
GBM
and
S.
cerevisiae
Gle1
CTD
•IP6•Dbp5
∆
N90
(ADP)
(PDB
ID
3RRN)
structures
3
. (c)
Comparison
of the
C.
thermophilum
Gle1
CTD
•Nup42
GBM
•IP
6
and
S.
cerevisiae
Gle1
CTD
•IP
6
•Dbp5
∆
N90
(ADP)
structures
(PDB
ID
3RRN).
(d)
Zoomed
view
of
the
IP
6
binding
pocket
in
the
C.
thermophilum
Gle1
CTD
•Nup42
GBM
•IP
6
and
S.
cerevisiae
Gle1
CTD
•IP
6
•Dbp5
∆
N90
(ADP) (
PDB ID 3RRN) structures.
19
20