Regulation of Torsin ATPases by LAP1 and LULL1
Chenguang Zhao, Rebecca S. H. Brown, Anna R. Chase, Markus R. Eisele, and Christian Schlieker1
Department of Molecular Biophysics and Biochemistry, Yale University, New Haven, CT 06520-8114
Edited by Tom A. Rapoport, Harvard Medical School, Howard Hughes Medical Institute, Boston, MA, and approved March 7, 2013 (received for review
January 22, 2013)
TorsinA is a membrane-associated AAA+ (ATPases associated with
a variety of cellular activities) ATPase implicated in primary dysto-
nia, an autosomal-dominant movement disorder. We reconstituted
TorsinA and its cofactors in vitro and show that TorsinA does not
display ATPase activity in isolation; ATP hydrolysis is induced upon
association with LAP1 and LULL1, type II transmembrane proteins
residing in the nuclear envelope and endoplasmic reticulum. This
interaction requires TorsinA to be in the ATP-bound state, and can
be attributed to the luminal domains of LAP1 and LULL1. This
ATPase activator function controls the activities of other members
of the Torsin family in distinct fashion, leading to an acceleration
of the hydrolysis step by up to two orders of magnitude. The
dystonia-causing mutant of TorsinA is defective in this activa-
tion mechanism, suggesting a loss-of-function mechanism for this
DYT1 dystonia|LINC complex|nuclear egress
(1). This lesion results in the loss of one of two consecutive glu-
tamate residues (E302 and E303) in TorsinA (TorA), a member
of the AAA+ (ATPases associated with a variety of cellular ac-
tivities) ATPases (2). Members of this superfamily typically as-
semble into ring-shaped, often hexameric, assemblies and use the
energy of ATP hydrolysis to exert force on their substrates. In
many cases, this force is invested in unfolding of targeted poly-
peptides or disassembly of protein complexes (3).
Of the four torsins that are encoded in the human genome
(TorsinA, TorsinB, Torsin2A, Torsin3A), TorA is by far the best
characterized (for a recent review, see ref. 4). TorA is broadly
expressed in humans (1), and is more abundant in neuronal than
in nonneuronal tissues in mice (5). TorsinA-null mice exhibit
early postnatal lethality despite a lack of obvious developmental
defects (6). Neurons in the CNS of these mice display abnor-
malities of the nuclear envelope (NE), namely a build-up of
vesicular structures in the perinuclear space. A highly similar
phenotype was seen in “knock-in” Tor1AΔgag/Δgagmice where
the GAG deletion in the Tor1A gene was introduced into the
TorA-null background. That the dystonia-associated TorA allele
fails to rescue the phenotype of a TorA-deficient mouse model
thus suggests a loss-of-function mechanism (6).
What is known about the molecular properties and function of
TorA? TorA features an N-terminal signal sequence that directs
it into the lumen of the endoplasmic reticulum (ER), where the
signal sequence is removed and two N-glycans are installed (7, 8).
TorA is recruited to the luminal side of the membrane by virtue
of an N-terminal hydrophobic domain (Fig. S1A) (9, 10), which
additionally serves to exclude TorA from ER exit sites (11). WT
TorA is partitioned between the ER and NE, whereas the dystonia-
associated mutant (hereafter referred to as TorA ΔE) forms
inclusions in the NE (6, 7, 12–14). TorA “trap” mutants rendered
ATP hydrolysis-deficient by mutating a residue in the Walker B
motif (TorA E171Q) are concentrated at the NE, suggesting a
function for TorA in the NE (13, 14). Two type II transmembrane
proteins have been implicated in controlling the subcellular local-
ization of TorA: lamina-associated polypeptide (LAP1) binds to
the nuclear lamina in the nucleus (15), while recruiting TorA to the
NE (16). Similarly, LULL1, which features a LAP1-like luminal
arly-onset DYT1 dystonia is a severe movement disorder that
is caused by an in-frame GAG deletion in the TOR1A gene
domain (LD), associates with TorA as well (16). Whether these
interactions are direct or indirect is at present unknown. Of note,
a LAP1-deficient mouse model displays perinatal lethality. CNS
neurons of these animals feature NE abnormalities that phenocopy
those seen in Tor1A−/−or Tor1AΔgag/Δgaganimals (5), suggesting
that TorA and LAP1 might act in the same pathway. However, the
mechanistic basis and functional significance of these interactions
are currently not well understood. Indeed, a functional assignment
is lacking for the majority of the proteins that have been identified
as components of the NE (17, 18). At present, our mechanistic
understanding of TorA in the functional context of its binding
partners is hampered by the absence of a suitable in vitro system.
Here we report a functional in vitro reconstitution of TorA
with its NE-resident binding partners and elucidate a role for
LAP1 and LULL1 as regulatory cofactors that are responsible
for the activation of TorA’s dormant ATPase activity. Further-
more, we demonstrate that most but not all Torsins are regulated
via a similar mechanism and thus establish a unique and con-
served regulatory role for these proteins. This study extends our
view of NE constituents—often perceived as passive scaffolding
components—and provides a quantitative and direct dem-
onstration of a loss-of-function mechanism in the context of
Glutamate Residue Deleted in Primary Dystonia Contributes to a
Solvent-Exposed Acidic Patch on the Surface of TorsinA. In the absence
of structural information for TorsinA, we generated a structure
model comprising residues 30–327. A cartoon representation of
the TorA model is given in Fig. S1. Given that the AAA domains
of many experimentally determined structures are highly similar
(19), we expect our model to be fairly accurate. The glutamate
residues (E302/E303) affected in primary dystonia, shown as
spheres, are located in the C-terminal four-helix bundle of the
AAA domain. Both residues are predicted to be solvent exposed
and contribute to an acidic patch (Fig. S1C).
Torsins belong to the AAA+ (ATPases associated with a variety
of cellular activities) ATPase superfamily, the members of
which disassemble protein complexes or unfold proteins. Here,
we provide evidence that the activity of Torsins is tightly reg-
ulated by two proteins that reside in the endoplasmic re-
ticulum and the perinuclear space. This regulatory mechanism
provides the framework for a better understanding of pheno-
types in animal models, and allows us to define the molecular
defect underlying TorsinA dystonia.
Author contributions: C.Z. and C.S. designed research; C.Z., R.S.H.B., A.R.C., M.R.E., and
C.S. performed research; C.Z., R.S.H.B., A.R.C., M.R.E., and C.S. contributed new reagents/
analytic tools; C.Z., R.S.H.B., A.R.C., M.R.E., and C.S. analyzed data; and C.Z., R.S.H.B., A.R.C.,
and C.S. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| Published online April 8, 2013
TorA migrates in blue native polyacrylamide gels at a position
that suggests a hexameric oligomerization state (9). Thus, we
generated a hexameric model to assess how the glutamate residues
are positioned therein (Fig. S1 D and E). Although approximately
half of the N-terminal hydrophobic domain is not represented in
our model, the fact that all of the N termini are predicted to orient
toward one side of the structure leads us to assume that the
hexamer is oriented relative to the membrane as depicted in
Fig. S1E, allowing for immersion of the hydrophobic domains
on the upper face of the hexamer into the lipid bilayer. In this
orientation, E302/E303 would be positioned on the luminal face
of the hexamer, being accessible from the ER/NE lumen. Thus,
TorA interaction partners that are known to reside in the ER/NE
lumen are candidates to interact with TorsinA in a manner that is
sensitive to structural alterations imposed by the glutamate deletion.
LAP1 and LULL1 Associate with TorsinA via Their Luminal Domains in
Vivo. To determine which interactions of TorA are affected by the
dystonia-causing mutation, C-terminally HA-tagged TorA expres-
sion constructs, including TorA WT, TorA ΔE (specifically, TorA
ΔE302), and TorA E171Q, were cloned. The latter mutant was
chosen because it is predicted to be trapped in an ATP-bound
state and such ATP “trapped” mutants show a high affinity for
substrate or cofactor binding in related ATPases (3). We also
created a construct with both the TorA E171Q and TorA ΔE
mutations, designated TorA E171Q/ΔE. HEK 293T cells were
transfected with TorA WT or its mutant derivatives and met-
abolically labeled with35S methionine/cysteine. Mild detergent
extracts were prepared from transfectants 24 h posttransfection,
and subjected to immunoprecipitation with anti-HA antibodies
in the presence of 2 mM ATP. The resulting immunoprecipitates
were resolved by SDS/PAGE and visualized via autoradiography
(Fig. 1A). The most prominent interaction partner, which asso-
ciated with all TorA variants indiscriminately, was calnexin (Fig.
1A). Because this previously reported interaction (20) was in-
sensitive to the dystonia-causing mutation, we shall address its
Notably, a protein with an apparent molecular mass of ∼60 kDa
was uniquely present in the immunoprecipitate derived from TorA
E171Q cells (Fig. 1A). This observation immediately suggests that
this protein binds selectively to the ATP-bound state of TorsinA.
Importantly, this interaction was not seen in the additional pres-
ence of the ΔE mutation (Fig. 1A). Thus, this interaction partner
is a prime candidate for further investigation in relation to primary
and sufficient for interaction with TorsinA in vivo.
(A) An ATP arrested TorsinA mutant associates with
LULL1. (Left) 293T cells were transfected with the
indicated constructs, metabolically labeled with 35S
methionine, and lysed in mild detergent 24 h post-
transfection. Following immunoprecipitation with
anti-HA antibodies, eluates were resolved by SDS/
PAGE and visualized by autoradiography. CNX, cal-
nexin. (Right) An immunoprecipitate obtained from
TorAE171Q transfectants was eluted with SDS, diluted
in buffer, and reimmunoprecipitated successively using
anti-LULL1 and anti-TorA antibodies. The resulting
immunoprecipitates were subjected to SDS/PAGE and
autoradiography. Fifty percent of the eluate used for
immunoprecipitation was loaded for comparison (in-
put, left lane). (B) TorA E171Q associates with LAP1.
(Left) 293T cells were transfected with the indicated
constructs and subjected to an immunoprecipitation
experiment as described in A. (Right) A reimmuno-
precipitation experiment was performed as described
above, using an anti-HA immunoprecipitate obtained
from a TorA E171Q/FLAG-LAP1 double transfectant
as starting material. Anti-TorA and anti-LAP1 anti-
sera were used for reimmunoprecipitation. (C and D)
Schematic view of LULL1 (C) and LAP1 (D) constructs
used in this study. TM, transmembrane domain; SS,
signal sequence. (E and F) The LDs of LULL1 and LAP1
associate with TorsinA. 293T cells were transfected
with the indicated constructs, and subjected to mild
detergent lysis and anti-HA immunoprecipitation. In-
put controls and immunoprecipitates were resolved
by SDS/PAGE and subjected to immunoblotting using
the indicated antibodies.
The LDs of LAP1 and LULL1 are necessary
| www.pnas.org/cgi/doi/10.1073/pnas.1300676110Zhao et al.
dystonia. Given that the apparent molecular mass of ∼60 kDa is
in the range of LULL1 (theoretical molecular mass = 51.3 kDa),
a known TorA interaction partner (16, 20), we repeated this
experiment in cells that were pretransfected with a LULL1
siRNA pool. Under these conditions, the 60-kDa band is no longer
observed, suggesting that this species is indeed LULL1. This in-
terpretation was confirmed by dissociating an anti-HA immuno-
precipitate obtained from TorA E171Q transfectants, followed
by reimmunoprecipitation of the eluate using an anti-LULL1 anti-
serum: this antiserum selectively retrieved the 60-kDa band that
we ascribed to LULL1 (Fig. 1A, Right).
We next tested if the related LAP1 protein shares the binding
characteristics of LULL1. FLAG-tagged LAP1 was cotransfected
with HA-tagged TorA WT and mutant derivatives, and subjected
to anti-HA immunoprecipitation after metabolic labeling. Follow-
ing SDS/PAGE and autoradiography, an additional band with the
expected molecular mass of FLAG-LAP1 (67.3 kDa) was indeed
observed that was absent from controls (Fig. 1B). To confirm the
proposed identity of this band, we dissociated immunoprecipitates
obtained from a duplicate FLAG-LAP1/TorA E171Q transfection,
and subjected them to a secondary immunoprecipitation using an
anti-LAP1 antiserum. The resulting autoradiogram unambiguously
confirms our previous interpretation: the electrophoretic mobility
of the obtained species corresponds exactly to the band ascribed
to LAP1 (Fig. 1B, Right). That endogenous LAP1 was not seen
in the autoradiogram we attribute to its inefficient solubilization
from the nuclear lamina at the moderate salt concentrations (20)
used in our experiments.
We next investigated which domains of LAP1 and LULL1 are
responsible for the interaction. The most likely candidates are the
LDs (see Fig. 1 C and D for domain architecture). This assumption
derives support from the observation that the ectopic expression
of LAP1/LULL1 constructs that encompass the LDs change the
subcellular localization of TorA, but constructs that lack them do
not (16). Because there is, to our knowledge, no direct evidence
for a physical interaction between the LDs of LAP1 or LULL1
and TorA, we engineered expression constructs of these domains
by installing a cleavable signal sequence on the N-terminal bound-
ary of the LD, and an HA tag at the C terminus. The corre-
sponding proteins, designated LAP1LDand LULL1LD, are
correctly targeted to the ER lumen, as judged by immunofluo-
rescence microscopy (Fig. S2).
To determine whether LAP1LDand LULL1LDindeed associate
with TorA, we cotransfected either construct with TorA E171Q
or a vector control, prepared detergent extracts 24 h post-
transfection, and subjected them to immunoprecipitation using
anti-HA antibodies. Immunoprecipitates were resolved by SDS/
PAGE and analyzed by immunoblotting using an anti-TorA
antiserum. TorsinA was only detectable in immunoprecipitates
obtained from LAP1LD- or LULL1LD-expressing cells (Fig. 1 E
and F). To exclude the formal possibility that the N-terminal
cytosolic/nuclear domains (CNDs) of LAP1 and LULL1 con-
tribute to the interaction indirectly, we also cloned these CNDs,
equipped with an N-terminal HA tag and a C-terminal farnesylation
signal for membrane recruitment. As expected, an interaction
between TorA and the CNDs was not observed (Fig. 1 E and F).
We conclude that LAP1 and LULL1 associate with TorsinA
preferentially in its ATP-bound form in a manner that is sensitive
to the dystonia-causing mutation. We additionally demonstrate
that the LDs of these membrane proteins are necessary and suf-
ficient for this interaction, which may be direct or indirect.
LAP1 and LULL1 Associate with TorsinA via Their LDs in Vitro. To
unambiguously establish whether the interaction between LULL1/
LAP1 and TorA is direct or indirect, we set up in vitro experi-
ments using purified components. Because full-length TorA bears
an N-terminal hydrophobic domain of established importance
(Fig. S1A) (11), we chose to express full-length TorA and its
mutant derivatives in a baculovirus system. We therefore in-
cluded the mild detergent dodecylmaltoside (DDM) throughout
the purification and all biochemical experiments described here-
after (Methods). TorA WT and its mutant derivatives were purified
to ∼95% homogeneity (Fig. 2A). The complex banding pattern
of purified TorA was reduced to one band by addition of Peptide
N-Glycanase F (PNGaseF), indicating that the apparent hetero-
geneity is attributable to incomplete N-glycosylation (Fig. 2A).
The LDs of LAP1 and LULL1 were expressed in Escherichia coli
and purified to apparent homogeneity (Fig. 2A).
We started our analysis by subjecting LAP1LDand LULL1LD
to size-exclusion chromatography. Based on the UV absorption
trace, both proteins eluted between the 44-kDa and 17-kDa marker
proteins of a size standard used for calibration (Fig. 2B). This
finding is consistent with a predicted molecular mass of 26.6 kDa
and 27.1 kDa for LAP1LDand LULL1LD, respectively. We con-
clude that LAP1LDand LULL1LDare monomeric under the tested
conditions. We additionally confirmed the identity of the UV peaks
by subjecting the corresponding elution fractions to immunoblot-
ting, using antisera raised against LAP1LDand LULL1LD(Fig. 2B).
We next tested whether the elution profile of LAP1LDor
LULL1LDis shifted to the high-molecular mass range in the
presence of TorA E171Q, which would be indicative of a direct
interaction. TorA E171Q was incubated with a twofold (monomer:
monomer) excess of LAP1LDor LULL1LDin the presence of
2 mM ATP and subjected to a gel-filtration column that was
preequilibrated in 500 μM ATP. A substantial shift toward the
high molecular-mass range was observed both for LAP1LDor
LULL1LD, as judged by immunoblotting of the resulting fractions
(Fig. 2 B and C). This effect was more pronounced for LULL1LD,
suggesting that LULL1 has a higher affinity for TorsinA. Corre-
spondingly, the absorption maximum of the peak corresponding to
free (monomeric) LAP1LDor LULL1LDwas reduced (Fig. 2C).
To establish whether ATP is required for this interaction, we
repeated these experiments but omitted the nucleotide both in the
binding reaction and the gel-filtration buffer. Under these con-
ditions, the shift of the elution profiles of LAP1LDand LULL1LD
upon addition of TorA E171Q was greatly reduced (Fig. 2D). This
finding was again confirmed by immunoblotting of the fractions
(Fig. 2D). In all, these data substantiate our earlier proposal that
it is in its ATP bound state that TorA has a high affinity to LAP1
or LULL1 (for additional controls, see Fig. S3).
LAP1 and LULL1 Do Not Efficiently Interact with TorA E171Q/ΔE in
Vitro. We next investigated whether the dystonia-associated glu-
tamate deletion affects LAP1 or LULL1 binding. TorA E171Q/
ΔE was incubated with LAP1LDor LULL1LDin presence of 2 mM
ATP and subjected to a gel-filtration column that was preequili-
brated in 500 μM ATP, as described above for the TorA E171Q
mutant. We did not observe a significant shift to an earlier elution
volume for either LD, as judged by immunoblotting of the ob-
tained fractions (Fig. 2E). We conclude that the interaction of
TorA E171Q/ΔE with both LDs is significantly impaired rela-
tive to the TorA E171Q trap mutant, which binds both domains
tightly. Thus, the in vitro data (Fig. 2) are in perfect agreement
with our binding studies in a cellular context (Fig. 1).
LDs of LAP1 and LULL1 Induce the ATPase Activity of TorsinA. What
are the functional consequences of LAP1 and LULL1 association
with TorsinA? We first tested whether LAP1 or LULL1 affect the
ATPase activity of TorA. Three micromolars of TorA WT was
incubated in the presence of 2 mM ATP, and phosphate release
was monitored at an endpoint after 60-min incubation. Surprisingly,
phosphate release in the presence of TorA WT did not differ from
background controls, indicating that TorsinA alone is catalytically
inactive (Fig. 3A).
Next, TorA WT was incubated in the additional presence of
3 μM LAP1LDor LULL1LD. Both proteins potently induced the
Zhao et al.PNAS
| Published online April 8, 2013
ATPase activity, with LULL1LDbeing more efficient than LAP1LD
at identical concentrations (Fig. 3A). This observation suggests
that LAP1 and LULL1 are positive regulators of TorsinA’s
To rule out that the observed ATPase activity is attributable
to a contaminating ATPase that was inadvertently copurified, we
additionally tested the TorA E171Q mutant. Analogous muta-
tions in related AAA+ ATPases render them catalytically in-
active (ref. 3 and references therein). The fact that TorA E171Q
does not display ATPase activity alone or in combination with
LAP1LDor LULL1LDexcludes this possibility (Fig. 3A).
Finally, we tested TorA ΔE. TorA ΔE did not display activity
in absence of LAP1LDor LULL1LD, as was observed for TorA
WT. More importantly, TorA ΔE activity failed to respond to
addition of LAP1 and LULL1 (Fig. 3A). This finding therefore
correlates the defect in LAP1/LULL1 binding (Figs. 1 and 2) with
a lack of ATPase induction (Fig. 3A). It seems reasonable to
propose that this molecular defect is a major contributor to the
etiology of primary dystonia.
LAP1 and LULL1 Have a Similar Mode of Action. The LDs of LAP1
and LULL1 are 60% identical on the level of primary structure,
suggesting that they operate via a similar mechanism. If these
cofactors would, however, accelerate distinct steps of the ATPase
cycle (e.g., ATP hydrolysis or nucleotide dissociation), we would
expect to see synergy between the two when added simultaneously.
However, an equimolar mixture of LAP1 and LULL1 (each
1.5 μM) stimulates ATP hydrolysis to a level that closely matches
with TorsinA is direct and ATP-dependent. (A, Left)
Purified LDs of LAP1 and LULL1 along with TorsinA
or indicated mutant derivatives were subjected to
SDS/PAGE and colloidal blue staining. (Right) Puri-
fied TorA WT was deglycosylated by incubation with
PNGase F and subjected to SDS/PAGE and immuno-
blotting using anti-TorsinA antibodies. (B–E) Protein
complex formation was monitored by size-exclusion
chromatography. UV traces are shown in blue, elu-
tion positions of size markers are indicated by arrows
on top. Elution fractions were subjected to immu-
noblotting using the indicated antibodies. (B) LDs of
LAP1 or LULL1 were subjected to size-exclusion chro-
matography. (C) TorA E171Q was incubated with a
twofold molar excess of LAP1LDor LULL1LDin the
presence of 2 mM ATP and subjected to a Superdex
200 PC 3.2/30 column preequilibrated in 500 μM ATP.
(D) TorA E171Q was incubated with a twofold molar
excess of LAP1LDor LULL1LDin the absence of ATP
during incubation or chromatographic separation.
(E) TorA E171Q/ΔE was incubated with a twofold
molar excess of LAP1LDor LULL1LDin presence of 2 mM
ATP and analyzed as in B. Note that additional controls
are shown in the supplement (Fig. S3).
The physical association of LAP1 and LULL1
| www.pnas.org/cgi/doi/10.1073/pnas.1300676110Zhao et al.
the arithmetic mean between the separate rates obtained for 3.0 μM
LAP1 or LULL1 added individually (Fig. 3B); that is, a synergistic
effect was not observed. This finding argues against the pos-
sibility that LAP1 and LULL1 accelerate distinct steps of the
LULL1 Is a More Potent Inducer of TorsinA ATPase Activity than LAP1.
Next, we determined kinetic parameters of the ATPase reaction.
We varied the concentration of LAP1 or LULL1 while leaving
the TorsinA concentration constant. We observed hyperbolic
saturation curves both for LAP1 and LULL1 titrations (Fig. 3 C
and D). The data were fitted to Michaelis–Menten kinetics and
yielded maximal velocities of 0.075 ± 0.007 μM·min−1and 0.14 ±
0.005 μM·min−1for LAP1 and LULL1, respectively. These values
correspond to turnover numbers of 0.16 min−1and 0.47 min−1
relative to a TorA monomer. Half-maximal stimulation was seen
in presence of 1.93 μM LAP1 or 0.65 μM LULL1, respectively.
Thus, LULL1 is a more potent inducer of TorsinA ATPase
activity than LAP1 for two reasons: (i) LULL1 has an approxi-
mately threefold-higher affinity for TorA than LAP1 (this is only
an approximation based on apparent Kmvalues), and (ii) LULL1
stimulates the ATPase to an approximately threefold higher Vmax.
We next assessed the stoichiometry of LAP1/LULL1 binding
via Job plot analysis (21). To this end, we measured the ATPase
activity in the presence of varying LAP1 or LULL1:TorA ra-
tios while leaving the sum of the concentrations of TorA and
cofactor constant. For LAP1, maximum intensity was observed
at a ratio that is very close to 1:1 (LAP1 monomer:TorA mono-
mer) (Fig. 3E). For LULL1, maximum ATPase activity was
seen at a ratio of ∼0.26–0.28:1 (LULL1 monomer:TorA mono-
or in presence of the LDs of LAP1LDor LULL1LD. Pi production as measure of ATP hydrolysis was monitored after 60 min using a malachite green assay. (B)
Next, 3 μM TorA was incubated at with 3 μM LAP1LDor LULL1LDindividually or a combination of both 1.5 μM LAP1LDand 1.5 μM LULL1LD. Pi release was
monitored after 60 min. (C and D) Initial velocities of ATP hydrolysis were obtained after monitoring Pi release in presence of 3 μM TorsinA and increasing
LAP1LD(C) or LULL1LD(D) concentrations as indicated. To increase the signal, the assay for LAP1LDwas performed in 40 μL instead of 25 μL. The data were
fitted to Michaelis–Menten kinetics in Prism and yielded the indicated apparent Kmand Vmaxvalues. (E and F) Job plot analysis to assess the stoichiometry of
LAP1LD/LULL1LDbinding to TorA. The total concentration of LAP1LD(or LULL1LD) and TorA in the ATPase assay was kept constant at 10 μM while varying the
ratios of individual components.
LAP1 and LULL1 stimulate the ATPase activity of TorsinA. (A) TorsinA and its mutant derivatives were incubated in presence of 2 mM ATP, either alone
Zhao et al.PNAS
| Published online April 8, 2013
mer) (Fig. 3F). Thus, LULL1 and LAP1 bind to TorA with distinct
Reconstitution of Torsin A in Proteoliposomes.Given that the ATPase
activity of TorA is low compared with other ATPases even in the
presence of LAP1 or LULL1, we tested whether lipids have a
stimulatory effect on activity, as is often seen for membrane pro-
teins. To this end, TorA was incubated with a mixture of lipids that
resembles the composition of ER membranes (22, 23), either alone
or in the additional presence of LAP1 or LULL1. The ATPase
activity in either case was comparable to control reactions that
were preformed with DDM-solubilized TorA (Fig. 4A).
We next reconstituted TorA in proteoliposomes to more closely
mimic its physiologically relevant environment in apposition to a
lipid bilayer. To assess the orientation of TorA in these proteoli-
posomes, we performed a protease protection assay. The majority
of TorA was readily degraded upon addition of proteinase K, even
in the absence of detergent, indicating that TorA preferentially
adopts an outside-facing orientation on proteoliposomes (Fig.
4B). In this orientation, TorA is accessible to exogenously added
nucleotides or cofactors. A small percentage of TorA was protease-
resistant unless Nonidet P-40 was additionally added, suggesting
that a minor fraction of TorA was luminal.
We then compared the ATPase activity of these proteolipo-
somes with free (i.e., DDM-solubilized) TorA under conditions
where the total concentration of TorA in both samples is identical
(Fig. 4C). In a direct comparison, the activity of TorA in pro-
teoliposomes is slightly lower than for free TorA, both in the
presence of LAP1 and LULL1 (Fig. 4A). We speculate that the
somewhat lower activity is attributable to a loss of biological ac-
tivity during the reconstitution process and because of the fact that
a small portion of TorA is luminal, and therefore inaccessible to
LAP1 and LULL1.
Taken together, these data demonstrate that the low ATPase
activity of TorA cannot be attributed to the lack of lipids that
would be present in a physiological setting.
Are LAP1 and LULL1 Substrates or Cofactors?For many AAA ATPases
that act to unfold substrates, the substrate itself can stimulate ATP
hydrolysis. It is therefore unclear whether LAP1 and LULL1 are
bona fide cofactors or merely substrates. We therefore tested
whether LAP1 or LULL1 are unfolded in the presence of TorA.
We set up a system that allows us to capture unfolded species using
a GroEL D87K “trap” variant, which binds to unfolded species
with high affinity, resulting in their sequestration into the central
GroEL cavity (24). The underlying rationale is that any unfolded
LAP1 or LULL1 species that would be produced by TorA’s
putative unfolding activity will immediately be sequestered by
the GroEL trap (present in excess), which is easily separated from
native species via size-exclusion chromatography (25, 26).
To validate our approach, we included positive controls in which
we chemically denatured LAP1 or LULL1 using 8 M urea, followed
by rapid dilution into a reaction mixture that was then chro-
matographically separated. Under these conditions, we can ro-
bustly detect unfolded LAP1/LULL1 species in association with
the GroEL trap, as judged by a profound shift of LAP1/LULL1
to GroEL-containing fractions that were subjected to immuno-
blotting with anti-LAP1/anti-LULL1 antisera (Fig. 5). Thus, we
would easily detect if even a small fraction of LAP1/LULL1 would
be unfolded by TorA. However, we failed to detect any evidence
for an unfolding activity, as judged by a complete lack of GroEL
trap-associated LAP1/LULL1 species in presence of TorA and an
ATP-regenerating system. We therefore consider a LAP1/LULL1-
directed threading/unfolding activity for TorA unlikely. However,
we cannot formally exclude that LAP1/LULL1 are substrates in
an in vivo setting.
LAP1 and LULL1 Accelerate the Hydrolysis Step of the ATPase Cycle.
The unfolding experiment argues against but does not exclude a
possible role for LAP1/LULL1 as substrates. We therefore tested
whether LAP1/LULL1 act as ATPase activators. Most commonly,
GTPase or ATPase activators function by accelerating nucleotide
hydrolysis or by modulating nucleotide affinity. We exploited the
fact that TorA’s ATPase activity is negligible in the absence of
cofactors and isolated complexes of TorA and α-32P ATP via
rapid (∼2 min) gel filtration at 4 °C. We used the resulting com-
plexes to perform single turnover assays in which we can simul-
taneously determine α-32P ADP and α-32P ATP concentrations
at a time of choice after separation via TLC, providing a direct
readout for the ATP hydrolysis step.
In agreement with our steady-state measurements (Fig. 3),
ATP hydrolysis is negligible in the absence of cofactors (Fig. 6A).
We then added either cofactor to a final concentration of 10 μM
tivity was measured for detergent-solubilized and reconstituted TorA. ER
lipids (Methods) were added to detergent-solubilized TorA to a final con-
centration of 300 μM. (B) Proteoliposomes were incubated with increasing
concentrations of proteinase K (PNK), in the absence or presence of Nonidet
P-40, and subjected to SDS/PAGE and immunoblotting. (C) The amount of
detergent-solubilized TorA and TorA in the proteoliposomes used in above
ATPase assay was compared via SDS/PAGE and colloidal blue staining.
Reconstitution of TorsinA in proteoliposomes. (A) The ATPase ac-
| www.pnas.org/cgi/doi/10.1073/pnas.1300676110Zhao et al.
that was found to be saturating under steady-state conditions (Fig.
3 C and D). ATP hydrolysis is rapidly induced upon addition of
LAP1 and, consistent with our observations under steady-state
conditions, more potently upon addition of LULL1 (Fig. 6A).
The fact that TorsinA and TorsinB are 68% identical in pri-
mary structure, as well as functionally redundant in murine cells
(5), prompted us to investigate whether TorsinB is similarly ac-
tivated by LAP1 or LULL1. This was indeed the case (Fig. 6B).
The resulting rate constants (Fig. 6C) inform us that for both TorA
and TorB, the ATP hydrolysis step is strongly accelerated by either
cofactor, up to almost two orders of magnitude in the case of TorB/
LULL1 (Fig. 6D).
This is strong evidence that it is specifically the hydrolysis step
that is tightly regulated by LAP1 and LULL1. The observed mag-
nitude of stimulation provides a gratifying mechanistic explanation
for our data obtained under steady-state conditions, specifically
(i) the negligible activity in absence of cofactors (Fig. 3A) and
(ii) the lack of synergy between LAP1 and LULL1 (Fig. 3B).
To our knowledge, this report of an AAA+ ATPase whose hy-
drolysis step is subject to tight regulation via bona fide ATPase
activators is unique.
Distinct Regulation of Torsins by Their Cofactors. We next inves-
tigated whether the remaining representatives of the Torsin
family are similarly regulated. We purified Torsin2A and Torsin3A
and measured their ATPase activities. Either ATPase alone
was essentially inactive (Fig. 7), as previously observed for
Torsin A and B (Figs. 3A and 6 A and B). Surprisingly, Tor2A
was not significantly stimulated upon addition of either LAP1
LULL1LDwere incubated in presence of GroEL D87K and an ATP-regenerat-
ing system, or in the additional presence of TorA. After 30 min, the reaction
mixtures were applied onto a Superdex 200 PC column and eluted with trap
buffer. Fractions were collected and separated by SDS/PAGE, and subjected
to immunoblotting using antisera against LAP1 or LULL1. The UV elution
profile of GroEL D87K and the positions of size markers are included for
reference. As positive control, LAP1LDor LULL1LDwere denatured in 8 M
urea, and rapidly diluted 1:50 into otherwise identical reactions.
LAP1 and LULL1 are not unfolded by TorsinA. Native LAP1LDor
the ATPase cycle. (A and B) Single turnover kinetics of TorA/B ATPases alone
(basal rate) or in presence of the indicated cofactor. Data represent a mean
obtained from three independent experiments. Error bars indicate the SD.
(C) Rate constants obtained from fitting the data above to a single expo-
nential decay function, using Prism. (D) Stimulation of TorA/B by LAP1LD/
LULL1LDrelative to the basal rate as observed in A and B.
LAP1LDand LULL1LDspecifically accelerate the ATP hydrolysis step of
Zhao et al. PNAS
| Published online April 8, 2013
or LULL1, whereas Tor3A was only stimulated by LULL1 but
not by LAP1 (Fig. 7).
Thus, the LAP1/LULL1-mediated activation mechanism that
we identified for TorA is conserved for TorB, with the latter being
the more potent ATPase (Figs. 6C and 7). Surprisingly, Tor2A
does not effectively respond to either cofactor, and Tor3A is only
activated by LULL1.
How the activity of Torsin ATPases is regulated is currently un-
known. LAP1 and LULL1 are NE components that are found in
association with TorA (16, 20). Overexpression of these factors
profoundly changed the subcellular localization of TorA or specific
mutant derivatives, suggesting an important role for these cofactors
(9, 16). Furthermore, a LAP1-deficient mouse model phenocopies
certain features of TorA-deficient animals (5). Taken together,
these observations prompted us to investigate the mechanistic
relationship between LAP1, LULL1, and TorsinA.
We showed that TorA physically associates with LAP1 and
LULL1. This interaction is direct (Fig. 2) and is mediated by the
LDs of these related type II transmembrane proteins (Figs. 1 E
and F, and 2). Only TorA E171Q, a hydrolysis-deficient trap
mutant of TorA, binds to LAP1 and LULL1 with high affinity, as
judged by coimmunoprecipitation experiments (Fig. 1 A and B),
and does so in nucleotide-dependent fashion (Fig. 2 C and D).
This interaction is interrupted by the DYT1 dystonia-causing mu-
tation, ΔE (Figs. 1 A and B, and 2E), again suggesting a functionally
relevant relationship between LAP1/LULL1 and TorsinA.
What are the functions of LAP1 and LULL1? Having demon-
strated that the LDs are necessary and sufficient for TorA binding
(Fig. 1 E and F), we purified these domains along with full-length
TorA and asked whether their presence or absence has an impact
on the ATPase activity of TorsinA. We found that TorA alone is
essentially inactive (Figs. 3A and 6A). However, ATPase activity
was induced in the presence of LAP1 or LULL1, and activity in-
creased with addition of LAP1 and LULL1 in a dose-dependent
fashion. LULL1 was significantly more efficient at accelerating
ATPase activity than LAP1, as judged by an approximately
threefold lower apparent Kmand approximately threefold higher
Vmax, respectively (Fig. 3 C and D). Strikingly, TorA ΔE failed to
respond to either LAP1 or LULL1, even at concentrations that
are saturating for TorA WT (Fig. 3A).
Taking into account our single turnover measurements (Fig. 6),
our observations are consistent with a model where LAP1 or
LULL1 bind to TorA in the ATP-bound state to trigger ATP
hydrolysis. Although our data do not rule out a possible role as
substrates, we strongly favor a regulatory function for the follow-
ing reasons: First, LULL1 and LAP1 affect the subcellular local-
ization of specific TorA variants, whereas TorA did not have an
impact on LULL1 or LAP1 localization (9, 16). Second, TorA
did not display a LAP1/LULL1-directed unfolding activity (Fig.
5). Apart from activating Torsin ATPases, LAP1 or LULL1 may
additionally serve as substrate adaptors, although this additional
role is at present purely speculative. It will be interesting to study
how the differential stimulation of Torsins by their cofactors
(Fig. 7) correlates with biological activity, and to determine how
the binding of cofactors is relayed to the ATPase active site.
The fact that Tor2A associates with LAP1 (5) despite being un-
responsive to its stimulation (Fig. 7) suggests a rather complex
mechanistic relationship for these interactions, the structural
underpinnings and biological relevance of which remain to be
It is noteworthy that the ATPase activity of all Torsins is
comparatively low even in the presence of saturating cofactor
concentrations (Figs. 3, 6D, and 7). This finding we attribute to
the presence of a noncanonical Walker A motif that is conserved
among all Torsins (i.e., GxxxxGKN rather then GxxxxGKS/T).
Indeed, a Thr to Asn mutation in the Walker A motif of the
homologous AAA ATPase ClpB (to mimic the Walker A motif
of TorA) substantially reduces its ATPase activity (27). Never-
theless, it will be interesting to test whether substrates additionally
stimulate the activity of their cognate Torsin ATPases. α-Casein,
a constitutively misfolded protein that stimulates a host of AAA+
ATPases implicated in protein quality control, including other
membrane-associated representatives (28, 29), did not stimu-
late the activity of TorA (Fig. S4). This property, as well as the
noncanonical Walker A motif, sets Torsins apart from the ma-
jority of AAA+ ATPases that possess an unfoldase function that
is directed to misfolded proteins.
Our study provides a mechanistic rationale for many pheno-
typic observations that were made in the context of animal models.
First, the fact that similar phenotypes were observed in LAP1- and
TorA-deficient mice had previously suggested that LAP1- and TorA
act in one pathway (5): we propose that because of the lack of
LAP1, the TorA ATPase activity is essentially kept off in the NE.
Second, our findings explain why LAP1 deficiency results in a
stronger phenotype than TorA deficiency (5): LAP1 is required
both for TorA and TorB activation (Fig. 6B). We suggest that
TorB can partially compensate for the missing TorA function in
nonneuronal tissues in Tor1A−/−or Tor1AΔgag/Δgaganimals,
whereas a LAP1 knockout will essentially be equivalent to
a TorA/TorB double-knockout. It will be interesting to test this
prediction experimentally. Third, our data demonstrate a loss-
of-function mechanism for the TorA ΔE dystonia mutation,
thus explaining why a Tor1AΔgag/Δgag“knock-in” in mouse
model fails to rescue the Tor1A−/−phenotype (6).
How the formation of NE vesicles seen as part of the Tor1A−/−
or Lap1−/−phenotype is related to the TorA ATPase remains to
be established. NE dynamics are not limited to cell division, but
they also occur in interphase cells. Examples include the occurrence
of intraluminal vesicles in neurons (30) or during viral infection
(31, 32), and NE invaginations that reach into the nucleoplasm
(33). Given that AAA+ ATPases, such as NSF and VPS4, are
molars of each Torsin was incubated in presence of 2 mM ATP, either alone
or in presence of the LDs of LAP1 or LULL1. Pi production as measure of ATP
hydrolysis was monitored after 60 min using a malachite green assay.
Regulation of the Torsin family by LAP1 and LULL1. Three micro-
| www.pnas.org/cgi/doi/10.1073/pnas.1300676110 Zhao et al.
implicated in various forms of membrane dynamics (3, 34, 35),
one may speculate that NE-resident Torsin ATPases are prime
candidates to be relevant for membrane dynamics of the NE
(31, 36). So far no NE abnormalities have been observed upon
LULL1 manipulation (5, 9), suggesting that LAP1 is primarily
important for TorA function in the NE. Apparently, LULL1 and
LAP1 are not functionally redundant.
Whether TorA has a direct role in the formation of vesicular
structures (30, 31, 36) or whether their appearance is merely an
indirect consequence of TorA inactivation is at present unknown.
We believe that the many distinct functions that have been pro-
posed for TorA (ref. 4 and references cited therein) depend on the
association of specific cofactors that serve both to recruit this
ATPase to specific cellular sites and to activate its ATPase ac-
tivity. Given that TorA associates with components of the NE-
resident LINC (linker of nucleoskeleton and cytoskeleton) complex
and changes their localization (9, 37), it will be interesting to test
whether TorA plays a role in assembly or disassembly of these NE
bridges. Having established a robust in vitro system for studying
TorsinA and its regulatory cofactors, this work provides the basis
to further scrutinize TorsinA functions from a mechanistic per-
spective and deepens our understanding of molecular defects un-
derlying DYT1 dystonia.
Constructs for Mammalian Cell Expression, Cell Lines, and Transient Transfections.
All constructs were cloned into pcDNA3.1+vector using PCR-based standard
procedures. Full-length LAP1 and LULL1 constructs were cloned with an
N-terminal FLAG tag. LAP1 and LULL1 LDs (LAP1LDor LAP1356-583, and
LULL1LDor LULL1236-470) were cloned with an N-terminal MHC-I HLA-A
signal sequence and a C-terminal HA tag. LAP1 cytosolic/nuclear domain
(LAP11–339) and LULL1 cytosolic/nuclear domain (LULL11–215) were cloned
with an N-terminal HA tag and a C-terminal farnesylation sequence from
Lamin B1 (ASNRS-CAIM) (LAP1CND-Fand LULL1CND-F). HEK293T and HeLa cells
were grown in a humidified atmosphere and 5% (vol/vol) CO2in DMEM
supplemented with 10% (vol/vol) FBS. HEK293T cells were transfected at
∼80% confluence using Lipofectamine 2000 (Invitrogen), and HeLa cells were
transfected using X-tremeGENE DNA (Roche). Experiments were performed
24 h posttransfection unless otherwise specified.
Metabolic Labeling, Immunoprecipitation, and Immunoblotting. Twenty-four
hours posttransfection, HEK293T cells were metabolically labeled with 800 μCi
of [35S] methionine/cysteine (PerkinElmer) at 37 °C overnight. The cells were
washed twice with ice-cold PBS and lysed with lysis buffer (50 mM Tris, pH 7.5,
75 mM NaCl, 5 mM MgCl2, and 0.5% Nonidet P-40) supplemented with 2 mM
ATP and complete (Roche) protease inhibitors on ice for 10 min. After
centrifugation (20,000 × g, 4 °C, 10 min), supernatants were precleared
with Protein A agarose (RepliGen) for 1 h followed by immunoprecipita-
tion with HA affinity beads (Roche) at 4 °C for 3 h. Beads were washed
with lysis buffer four times, boiled in SDS loading buffer, and subjected to
SDS/PAGE. Immunoblotting was performed according to standard procedures,
using the Western Lightning Plus ECL chemiluminescence kit (PerkinElmer) and
X-ray films (Kodak) for detection.
Cloning, Bacterial Expression, and Purification of LAP1LDand LULL1LD. The
luminal segments of LAP1 (LAP1LDor LAP1356–583) and LULL1 (LULL1LDor
LAP1236–470) were cloned via PCR into a modified pET28b vector containing
a PreScission Protease restriction site between a His-tag and the fused pro-
tein. LULL1LDwas expressed in Origami 2(DE3)pLysS cells (Novagen). The cells
were grown at 30 °C in DYT medium [1.6% (wt/vol) tryptone, 1% yeast ex-
tract, and 0.5% NaCl, pH 7.0) supplemented with 50 μg/mL kanamycin and
34 μg/mL Chloramphenicol to OD600of 1.0 and protein production was in-
duced with 1 mM isopropyl-β-D-thiogalactopyranoside at 18 °C overnight.
The cells were harvested by centrifugation at 6,000 × g for 15 min and the
pellet was resuspended in binding buffer (20 mM sodium phosphate, 0.5 M
NaCl, 10 mM imidazole, pH 7.4) supplemented with Complete (Roche) pro-
tease inhibitors (Roche). Cells were disrupted using a microfluidizer at a pres-
sure of 15,000 psi. After removal of the insoluble fraction by centrifugation at
26,000 × g for 40 min, the supernatant was incubated with 1 mL Ni Sepharose
6 Fast Flow resin (GE Lifesciences) at 4 °C for 1 h. Unbound material was re-
moved with 30 mL binding buffer. His-tagged LULL1LDwas eluted with 5 mL
elution buffer (20 mM sodium phosphate, 0.5 M NaCl, 500 mM imidazole,
pH 7.4). The eluate was immediately diluted 10-fold with ion exchange
binding buffer (25 mM Tris and 50 mM NaCl, pH 7.5) and applied to
a Resource Q column (GE Lifesciences). The protein was eluted with a linear
gradient to 1 M NaCl in 20 column volumes, pooled and incubated with
PreScission protease (GE Lifesciences) (20 units/mg LULL1LD) overnight to
remove the His-tag. PreScission protease was removed by gel filtration using
a Superdex 75 HiLoad column (GE Lifesciences). For snap-freezing in liquid
nitrogen, 10% (vol/vol) glycerol was added to the protein solution.
LAP1LDwas expressed in Rosetta(DE3)pLysS cells (Novagen) at 30 °C and
purified as described for LULL1, except that the ion-exchange step was omitted.
Antibodies. Anti-FLAG antibody was purchased from Sigma-Aldrich and anti-
HA antibody (3F10) was purchased from Roche. Antibodies against LaminA
and LaminB1 were purchased from Abcam. Goat anti-rabbit, anti-mouse, and
anti-rat conjugated to Alexa Flour dyes were purchased from Invitrogen.
Goat anti-rabbit and anti-mouse IgG HRP conjugated antibodies were pur-
chased from Thermo Scientific. A TorA fragment lacking the N-terminal
signal sequence and hydrophobic segment was expressed in Rosseta(DE3)
pLysS cells (Novagen) and purified via His-tag for antiserum production
(Covance). The anti-LAP1 and anti-LULL1 antisera were raised against the
above purified LAP1LDand LULL1LD(Covance).
Cloning, Expression, and Purification of Human Torsins. Human Torsins and
its mutant derivatives were PCR-amplified to install a C-terminal His10-tag
followed by a Flag-tag and cloned into pFastBac I vector (Invitrogen). The
recombinant baculovirus generation and amplification were performed as
described by Invitrogen. Torsins were expressed for 72 h in Sf9 cells infected
with the recombinant baculovirus at a multiplicity of infection of 2. For
purification, the cells were centrifuged at 1,000 × g and 4 °C for 10 min and
the pellet was washed with ice-cold PBS (pH 7.4). The cells were solubilized
with lysis buffer [20 mM Hepes, 150 mM NaCl, 5 mM MgCl2, 5 mM KCl, 10%
(vol/vol) glycerol, and 1% (wt/vol) DDM, pH 8.0] supplemented with Com-
plete (Roche) protease inhibitors at 4 °C for 1 h and centrifuged at 20,000 × g
for 30 min. The supernatant was incubated with 1 mL Anti-FLAG M2 Affinity
Gel (Sigma) equilibrated with equilibration buffer (20 mM Hepes, 150 mM
NaCl, 5 mM MgCl2, 5 mM KCl, 10% glycerol, and 0.05% DDM, pH 8.0) at 4 °C
for 3 h followed by washing with 20 column volumes of equilibration buffer.
TorA was eluted by incubating with 5 column volumes of 300 μg/mL FLAG
peptide in equilibration buffer at 4 °C for 1 h. The eluate was subjected to
0.5 mL of His-Select Nickel Affinity Gel (Sigma) at 4 °C for 3 h in the presence
of 10 mM imidazole. After washing with 20 column volumes of equilibration
buffer (20 mM Hepes, 150 mM NaCl, 5 mM MgCl2, 5 mM KCl, 10% (vol/vol)
glycerol, and 0.05% DDM, pH 8.0) supplemented with 10 mM imidazole,
Torsins were eluted with 5 column volumes of 150 mM imidazole in equili-
bration buffer. Imidazole was removed by PD-10 column from GE Health-
care. For long-term storage, Torsins were snap-frozen in liquid nitrogen and
stored at –80 °C.
ATPase Activity Assay. For ATPase activity assay, 3 μM of purified TorA was
incubated in 25 μL equilibration buffer (20 mM Hepes, 150 mM NaCl, 5 mM
MgCl2, 5 mM KCl, 10% glycerol, and 0.05% DDM, pH 8.0) supplemented
with 2 mM ATP at 37 °C for 1 h, in the presence or absence of LAP1LDor
LULL1LD. The reaction was terminated with 175 μL of 20 mM sulfuric acid
(H2SO4). The concentration of inorganic phosphate (Pi) was determined
by a colorimetric method (38). Fifteen minutes after the addition of 50 μL
of freshly prepared malachite green solution [0.096% malachite green,
1.48% (wt/vol) ammonium molybdate, and 0.173% Tween-20 in 2.9 M H2SO4],
the absorbance was measured at 620 nm in a 96-well plate reader (BioTek).
Measurements were performed in triplicate under conditions giving a linear
rate of product formation (Pi). Calibration of free phosphate concentration
was carried out with Na2HPO4in 200 μL of 20 mM H2SO4. Data analysis was
performed in Prism.
For single turnover assays, 50 μM Torsin was incubated with 100 μCi of
α-32P (800 Ci/mmol; Perkin-Elmer) ATP for 3 min at 30 °C in a total volume of
100 μL and chilled on ice. Free ATP was separated from TorA-bound ATP by
a rapid gel filtration using a PD10 column (GE Healthcare) preequilibrated
in ice-cold running buffer (30 mM Hepes, 75 mM NaCl, 5 mM MgCl2, pH 7.5).
Complexes were detected in the eluate using a Geiger counter, pooled,
aliquoted on ice, and flash-frozen in liquid nitrogen. Single turnover assays
were performed at 30 °C. Two-microliter aliquots were removed at indicated
time points and resolved by TLC using PEI-Cellulose (Merck) and 400 mM LiCl
in 10% (vol/vol) acetic acid as mobile phase. TLC plates were imaged using
an imaging plate (GE Healthcare) and quantified densiometrically using
ImageQuant. The resulting data were fitted to a single exponential decay
function in Prism.
Zhao et al. PNAS
| Published online April 8, 2013
Analytical Gel Filtration. Purified TorA or mutant derivatives (5 μM) were
incubated with Lap1LDor LULL1LD(10 μM) in gel-filtration buffer [25 mM
Tris, 75 mM NaCl, 5 mM MgCl2, 5 mM KCl, 5% (vol/vol) glycerol, and 0.05%
(wt/vol) DDM, pH 7.5], in the presence or absence of 2 mM ATP at 30 °C for
5 min and subjected to a Superdex 200 PC 3.2/30 column on a Purifier system
(GE Healthcare). The protein was eluted with gel-filtration buffer in the pres-
ence or absence of 0.5 mM ATP, at a flow rate of 50 μL/min. The absorbances
at 280 nm were recorded. Fractions of 100 μL were collected and analyzed by
SDS/PAGE (12%) and immunoblotting. Soluble protein standards from Bio-Rad
were used for column calibration.
Proteoliposome Reconstitution. Reconstitution of TorA was performed as de-
scribed previously (39). In short, ER membrane lipids were prepared by mixing
cholesterol and phospholipids [55%(mol/mol) dioleoylphosphatidylcholine
(DOPC), 20% dioleoylphosphatidylethanolamine (DOPE), 5% dioleoylphos-
phatidylserine (DOPS), 5% dioleoylphosphatidic acid (DOPA), and 15%
phosphatidylinositol (PI)] in a molar ratio of 0.15. After removing the organic
solvent, the lipids were rehydrated in reconstitution buffer (30 mM Hepes,
100 mM NaCl, 5 mM MgCl2, pH 7.5). To prepare large unilaminar vesicles,
the liposomes were snap-frozen in liquid N2and thawed at room temper-
ature for five cycles followed by extrusion through a 400-nm polycarbonate
filter (Avestin). The large unilaminar vesicles were diluted to 2.5 mg/mL and
destabilized by 10% (wt/vol) Triton X-100, and followed by incubation with
TorA with a molar ratio of 700:1 (lipids to TorA) at 4 °C for 30 min. After
removal of detergents by addition of Bio-beads (Bio-Rad), the proteolipo-
somes were sedimented by centrifugation (300,000 × g for 45 min) and
resuspended in 450 μL reconstitution buffer.
GroEL Trap Assay. GroEL D87K (GroEL “trap,” plasmid kindly provided by Art
Horwich, Yale School of Medicine, New Haven, CT and Howard Hughes
Medical Institute) was purified as described previously (40). Lap1LDor
Lull1LDwas diluted to 0.3 μM in trap buffer (25 mM Tris, 150 mM NaCl, 5
mM MgCl2, 5 mM KCl, pH 7.5) containing 7 μM GroEL D87K, 2 mM ATP, 3
mM phosphoenolpyruvate and 1 U pyruvate kinase (PK) at 30 °C for 30 min,
in the presence or absence of 0.3 μM TorA. As a control, 8 M urea was in-
cubated with 50-fold LAP1LDor LULL1LDbefore a 1:50 dilution to a final
concentration of 300 nM in the unfolding reaction. The reaction mixture was
applied onto a Superdex 200 PC column and eluted with trap buffer.
ACKNOWLEDGMENTS. We thank members of the C.S. laboratory for comments
on the manuscript and Art Horwich for providing the GroEL D87K plasmid.
This work was funded by the Ellison Medical Foundation (AG-NS-0662-10)
and National Institutes of Health (DP2 OD008624-01).
1. Ozelius LJ, et al. (1997) The early-onset torsion dystonia gene (DYT1) encodes an
ATP-binding protein. Nat Genet 17(1):40–48.
2. Neuwald AF, Aravind L, Spouge JL, Koonin EV (1999) AAA+: A class of chaperone-like
ATPases associated with the assembly, operation, and disassembly of protein complexes.
Genome Res 9(1):27–43.
3. Hanson PI, Whiteheart SW (2005) AAA+ proteins: Have engine, will work. Nat Rev Mol
Cell Biol 6(7):519–529.
4. Atai NA, Ryan SD, Kothary R, Breakefield XO, Nery FC (2012) Untethering the nuclear
envelope and cytoskeleton: biologically distinct dystonias arising from a common
cellular dysfunction. Int J Cell Biol 2012:634214.
5. Kim CE, Perez A, Perkins G, Ellisman MH, Dauer WT (2010) A molecular mechanism
underlying the neural-specific defect in torsinA mutant mice. Proc Natl Acad Sci USA
6. Goodchild RE, Kim CE, Dauer WT (2005) Loss of the dystonia-associated protein
torsinA selectively disrupts the neuronal nuclear envelope. Neuron 48(6):923–932.
7. Hewett J, et al. (2000) Mutant torsinA, responsible for early-onset torsion dystonia,
forms membrane inclusions in cultured neural cells. Hum Mol Genet 9(9):1403–1413.
8. Callan AC, Bunning S, Jones OT, High S, Swanton E (2007) Biosynthesis of the dystonia-
associated AAA+ ATPase torsinA at the endoplasmic reticulum. Biochem J 401(2):
9. Vander Heyden AB, Naismith TV, Snapp EL, Hodzic D, Hanson PI (2009) LULL1 retargets
TorsinA to the nuclear envelope revealing an activity that is impaired by the DYT1
dystonia mutation. Mol Biol Cell 20(11):2661–2672.
10. Liu Z, Zolkiewska A, Zolkiewski M (2003) Characterization of human torsinA and its
dystonia-associated mutant form. Biochem J 374(Pt 1):117–122.
11. Vander Heyden AB, Naismith TV, Snapp EL, Hanson PI (2011) Static retention of the
lumenal monotopic membrane protein torsinA in the endoplasmic reticulum. EMBO J
12. Gonzalez-Alegre P, Paulson HL (2004) Aberrant cellular behavior of mutant torsinA
implicates nuclear envelope dysfunction in DYT1 dystonia. J Neurosci 24(11):2593–2601.
13. Goodchild RE, Dauer WT (2004) Mislocalization to the nuclear envelope: An effect of
the dystonia-causing torsinA mutation. Proc Natl Acad Sci USA 101(3):847–852.
14. Naismith TV, Heuser JE, Breakefield XO, Hanson PI (2004) TorsinA in the nuclear
envelope. Proc Natl Acad Sci USA 101(20):7612–7617.
15. Foisner R, Gerace L (1993) Integral membrane proteins of the nuclear envelope interact
with lamins and chromosomes, and binding is modulated by mitotic phosphorylation.
16. Goodchild RE, Dauer WT (2005) The AAA+ protein torsinA interacts with a conserved
domain present in LAP1 and a novel ER protein. J Cell Biol 168(6):855–862.
17. Schirmer EC, Florens L, Guan T, Yates JR, 3rd, Gerace L (2003) Nuclear membrane proteins
with potential disease links found by subtractive proteomics. Science 301(5638):
18. Hetzer MW, Wente SR (2009) Border control at the nucleus: Biogenesis and organization
of the nuclear membrane and pore complexes. Dev Cell 17(5):606–616.
19. Zhang X, et al. (2000) Structure of the AAA ATPase p97. Mol Cell 6(6):1473–1484.
20. Naismith TV, Dalal S, Hanson PI (2009) Interaction of torsinA with its major binding
partners is impaired by the dystonia-associated DeltaGAG deletion. J Biol Chem 284(41):
21. Cantor, CR, Schimmel PR (1980) Biophysical Chemistry: Part III (W. H. Freeman, San
Francisco), pp 1135–1139.
22. van Meer G, Voelker DR, Feigenson GW (2008) Membrane lipids: Where they are
and how they behave. Nat Rev Mol Cell Biol 9(2):112–124.
23. Davison SC, Wills ED (1974) Studies on the lipid composition of the rat liver endoplasmic
reticulum after induction with phenobarbitone and 20-methylcholanthrene. Biochem J
24. Weissman JS, Kashi Y, Fenton WA, Horwich AL (1994) GroEL-mediated protein folding
proceeds by multiple rounds of binding and release of nonnative forms. Cell 78(4):
25. Weber-Ban EU, Reid BG, Miranker AD, Horwich AL (1999) Global unfolding of a
substrate protein by the Hsp100 chaperone ClpA. Nature 401(6748):90–93.
26. Schlieker C, Tews I, Bukau B, Mogk A (2004) Solubilization of aggregated proteins by
ClpB/DnaK relies on the continuous extraction of unfolded polypeptides. FEBS Lett
27. Nagy M, Wu HC, Liu Z, Kedzierska-Mieszkowska S, Zolkiewski M (2009) Walker-A
threonine couples nucleotide occupancy with the chaperone activity of the AAA+
ATPase ClpB. Protein Sci 18(2):287–293.
28. Roudiak SG, Shrader TE (1998) Functional role of the N-terminal region of the Lon
protease from Mycobacterium smegmatis. Biochemistry 37(32):11255–11263.
29. Asahara Y, et al. (2000) FtsH recognizes proteins with unfolded structure and
hydrolyzes the carboxyl side of hydrophobic residues. J Biochem 127(5):931–937.
30. Speese SD, et al. (2012) Nuclear envelope budding enables large ribonucleoprotein
particle export during synaptic Wnt signaling. Cell 149(4):832–846.
31. Rose A, Schlieker C (2012) Alternative nuclear transport for cellular protein quality
control. Trends Cell Biol 22(10):509–514.
32. Johnson DC, Baines JD (2011) Herpesviruses remodel host membranes for virus egress.
Nat Rev Microbiol 9(5):382–394.
33. Malhas A, Goulbourne C, Vaux DJ (2011) The nucleoplasmic reticulum: Form and
function. Trends Cell Biol 21(6):362–373.
34. Hurley JH, Hanson PI (2010) Membrane budding and scission by the ESCRT machinery:
It’s all in the neck. Nat Rev Mol Cell Biol 11(8):556–566.
35. Henne WM, Buchkovich NJ, Emr SD (2011) The ESCRT pathway. Dev Cell 21(1):77–91.
36. Burns LT, Wente SR (2012) Trafficking to uncharted territory of the nuclear envelope.
Curr Opin Cell Biol 24(3):341–349.
37. Nery FC, et al. (2008) TorsinA binds the KASH domain of nesprins and participates in
linkage between nuclear envelope and cytoskeleton. J Cell Sci 121(Pt 20):3476–3486.
38. Baykov AA, Evtushenko OA, Avaeva SM (1988) A malachite green procedure for
orthophosphate determination and its use in alkaline phosphatase-based enzyme
immunoassay. Anal Biochem 171(2):266–270.
39. Zhao C, Haase W, Tampé R, Abele R (2008) Peptide specificity and lipid activation of
the lysosomal transport complex ABCB9 (TAPL). J Biol Chem 283(25):17083–17091.
40. Quaite-Randall E, Joachimiak A (2000) Purification of GroEL from an overproducing
E. coli strain. Methods Mol Biol 140:29–39.
| www.pnas.org/cgi/doi/10.1073/pnas.1300676110Zhao et al.
Zhao et al. 10.1073/pnas.1300676110
Torsin A Structure Prediction, Electrostatic Surface and Hexamer
Modeling. A TorsinA (TorA) model was created using Swissmodel
in automated mode (1), using ClpB (PDB entry 1QVR, chain C)
(2) as template. Predictions of electrostatic surface potential of the
TorsinA monomer were carried out using the San Diego Super-
computing Center online server PDB2PQR v1.8 (http://kryptonite.
nbcr.net/pdb2pqr/) (3) with PROPKA pKacalculation software at
pH 7.0 (4) and APBS Tools in PyMol (5). A hexameric model
of Torsin A was generated by superimposing its structure with
that of each monomer in the hexameric AAA (ATPases as-
sociated with a variety of cellular activities) ATPase p97 (PDB
entry 1R7R) (6) in Coot (7). All structure representations were
made using Pymol (5).
Immunofluorescence. For immunofluorescence, 25,000 HeLa cells
were seeded on a coverslip and transfected after 24 h using
X-tremGENE 9 (Roche). Immunofluroescence labeling was per-
PBS and fixed with 4% paraformaldehyde for 20 min at room
temperature. After permeabiliziation with 0.1% Triton X-100/
PBS for 10 min at room temperature, the coverslips were blocked
with 4% BSA/PBS for 20 min at room temperature. Coverslips
were incubated with primary antibody (1:500) in 4% BSA/PBS
overnight at room temperature. Coverslips were then washed
and blocked with 4% BSA/PBS for 20 min, and incubated with
Alexa488/568-conjugated secondary antibodies (Invitrogen) (1:700)
for 45 min in the dark. Following three washes in PBS, coverslips
were mounted using Fluoromount-G. Images were acquired using
a Zeiss microscope, model Observer D1, and 63×/1.4 oil lens.
1. Kiefer F, Arnold K, Kunzli M, Bordoli L, Schwede T (2009) The SWISS-MODEL Repository
and associated resources. Nucleic Acids Res 37(Database issue):D387–D392.
2. Lee S, et al. (2003) The structure of ClpB: a molecular chaperone that rescues proteins
from an aggregated state Cell 115(2):229–240.
3. Dolinsky TJ, et al. (2007) PDB2PQR: expanding and upgrading automated preparation
of biomolecular structures for molecular simulations. Nucleic Acids Res 35(Web Server
4. Li H, Robertson AD, Jensen JH (2005) Very fast empirical prediction and rationalization
of protein pKa values. Proteins 61(4):704–721.
5. Baker NA, Sept D, Joseph S, Holst MJ, McCammon JA (2001) Electrostatics of
nanosystems: application to microtubules and the ribosome. Proc Natl Acad Sci USA
6. Huyton T, et al. (2003) The crystal structure of murine p97/VCP at 3.6A. J Struct Biol
7. Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot.
Acta Crystallographica, Section D, Biological Crystallography 66(Pt 4):486–501.
code corresponds to the structure model depicted in B. (B) Cartoon representation of TorsinA. Glutamate residues implicated in primary dystonia are shown
in space fill representation. (C) Electrostatic surface potential representation of TorsinA with kT/e ± 1. (D) A hexameric model of TorsinA. Note that the hy-
drophobic domain, which is only partially represented in this model (see main text), would be appended to the N termini (four of six are highlighted), which
are facing upwards in the orientation shown in E.
A structural model for the TorsinA. (A) Domain organization of TorsinA. SS, cleavable signal sequence. Hy, hydrophobic domain. Note that the color
Zhao et al. www.pnas.org/cgi/content/short/13006761101 of 3
(lamina-associated polypeptide) and LULL1 (which features a LAP1-like luminal domain). HeLa cells were transfected with a vector control or the indicated
constructs depicted in Fig. 1 C and D. Twenty-four hours posttransfection, cells were processed for immunofluorescence microscopy using the indicated antibodies.
Cells were costained against LaminA or LaminB1 to mark the nuclear envelope.
LULL1LDand LAP1LD(LD, luminal domain) constructs localize to the endoplasmic reticulum (ER) lumen. (A and B) Subcellular localization of LAP1
Zhao et al. www.pnas.org/cgi/content/short/13006761102 of 3
TorA E171Q was incubated with 2 mM ATP and subjected to a Superdex 200 PC 3.2/30 column preequilibrated in 500 μM ATP. UV traces are shown in blue,
elution positions of size markers are indicated by arrows on top. Elution fractions were subjected to immunoblotting using the indicated antibodies. (B)
Chromatogram of TorA E171Q/ΔE in the absence (Upper) or presence (Lower) of ATP.
Additional size-exclusion chromatography controls for Fig. 2. (A) Chromatogram of TorA E171Q in the absence (Upper) or presence (Lower) of ATP.
the absence or presence of 3 μM Lap1LDor Lull1LDor mixture of Lap1LDand Lull1LD(each 1.5 μM) in the ATPase assay. For comparison, TorA ATPase activities
in the absence or presence cofactors were included.
TorA ATPase activity was not stimulated by α-casein. Proteoliposomes containing 3 μM TorA were incubated with 0.25 mg/mL (11 μM) α-casein, in
Zhao et al. www.pnas.org/cgi/content/short/13006761103 of 3