*For correspondence: tus@mit.
edu
Competing interest: See
page 11
Funding: See page 11
Received: 19 May 2016
Accepted: 03 August 2016
Published: 04 August 2016
Reviewing editor: Wesley I
Sundquist, University of Utah,
United States
Copyright Demircioglu et al.
This article is distributed under
the terms of the Creative
Commons Attribution License,
which permits unrestricted use
and redistribution provided that
the original author and source are
credited.
Structures of TorsinA and its disease-
mutant complexed with an activator
reveal the molecular basis for primary
dystonia
F Esra Demircioglu1, Brian A Sosa1, Jessica Ingram2, Hidde L Ploegh2,
Thomas U Schwartz1*
1Department of Biology, Massachusetts Institute of Technology, Cambridge, United
States; 2Whitehead Institute for Biomedical Research, Cambridge, United States
Abstract The most common cause of early onset primary dystonia, a neuromuscular disease, is a
glutamate deletion (DE) at position 302/303 of TorsinA, a AAA+ ATPase that resides in the
endoplasmic reticulum. While the function of TorsinA remains elusive, the DE mutation is known to
diminish binding of two TorsinA ATPase activators: lamina-associated protein 1 (LAP1) and its
paralog, luminal domain like LAP1 (LULL1). Using a nanobody as a crystallization chaperone, we
obtained a 1.4 A˚ crystal structure of human TorsinA in complex with LULL1. This nanobody likewise
stabilized the weakened TorsinADE-LULL1 interaction, which enabled us to solve its structure at 1.4
A˚ also. A comparison of these structures shows, in atomic detail, the subtle differences in activator
interactions that separate the healthy from the diseased state. This information may provide a
structural platform for drug development, as a small molecule that rescues TorsinADE could serve
as a cure for primary dystonia.
DOI: 10.7554/eLife.17983.001
Introduction
Torsins belong to the AAA+ (ATPases associated with a variety of cellular activities) ATPase family, a
functionally diverse group of enzymes, which are fueled by ATP hydrolysis. AAA+ ATPases organize
in structurally distinct fashions and interact with various accessory elements to remodel their protein
or nucleic acid substrates (Erzberger and Berger, 2006; Wendler et al., 2012; White and Lauring,
2007). Torsins are poorly understood AAA+ proteins with yet elusive functions and unknown sub-
strates (Laudermilch and Schlieker, 2016; Rose et al., 2015). Among the five human torsins (Tor-
sinA, TorsinB, Torsin2A, Torsin3A and Torsin4A), neuronally expressed TorsinA carries the most
clinical significance since it is at the root of primary dystonia. Primary dystonia is a devastating neuro-
muscular disease that is predominantly caused by the deletion of glutamate 302 or 303 (DE) in Tor-
sinA (Goodchild et al., 2005; Ozelius et al., 1997). The etiology of primary dystonia is poorly
understood (Breakefield et al., 2008; Granata and Warner, 2010), and there is currently no known
cure for it.
TorsinA is an unusual AAA+ ATPase, because, unlike any other family member (Erzberger and
Berger, 2006; Laudermilch and Schlieker, 2016; Rose et al., 2015; White and Lauring, 2007), it is
localized to the endoplasmic reticulum (ER) and the contiguous perinuclear space (PNS), and
because it is not self-activated, but instead needs the AAA+-like proteins Lamina-associated protein
1 (LAP1) or Luminal domain like LAP1 (LULL1) to catalyze ATP hydrolysis (Brown et al., 2014;
McCullough and Sundquist, 2014; Sosa et al., 2014). LAP1 is a type-II transmembrane protein,
which resides at the inner nuclear membrane (INM) through its association with the nuclear lamina
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 1 of 14
RESEARCH ARTICLE
(Goodchild and Dauer, 2005). LULL1 is a LAP1 paralog, which localizes to the outer nuclear mem-
brane (ONM) and the continuous ER, with its N-terminal portion protruding into the cytoplasm
(Goodchild and Dauer, 2005). The structurally similar luminal domains of LAP1/LULL1 interact with
TorsinA, and they provide an arginine finger to the TorsinA active site to facilitate torsin’s ATP
hydrolysis (Brown et al., 2014; Sosa et al., 2014). Arginine fingers are key structural motifs of AAA
+ ATPases because they neutralize the transition state during ATP hydrolysis (Wendler et al., 2012).
Since torsins lack arginine fingers themselves, this activation mechanism through LAP1/LULL1 is likely
critical for their function. As reported by several labs, the disease mutant TorsinA DE is compromised
in binding to LAP1/LULL1 (Naismith et al., 2009; Zhao et al., 2013; Zhu et al., 2010). Clearly, this
suggests that a probable cause of primary dystonia is the lack of activation of TorsinA. In line with
this suggestion, LAP1 deletion shows a similar phenotype to Torsin DE, and contributes to disease
pathology (Kim et al., 2010).
To investigate the molecular basis for primary dystonia as a result of the glutamate 302/303 dele-
tion in TorsinA, we took a structural approach. We obtained high-resolution crystal structures of Tor-
sinA as well as TorsinADE, each in complex with LULL1, using a nanobody as crystallization
chaperone. These structures likely open a pathway toward rational, structure-based drug design
against primary dystonia.
Results
TorsinA is a catalytically inactive AAA+ ATPase (Brown et al., 2014; Zhao et al., 2013), notoriously
ill-behaved in vitro, primarily due to its limited solubility and stability. We partially overcame these
problems by stabilizing an ATP-trapped E171Q mutant of human TorsinA (residues 51–332) by co-
expressing it with the luminal activation domain of human LULL1 (residues 233–470). This resulted in
a better behaved heterodimeric complex (Figure 1A), which, however, was still recalcitrant to our
crystallization efforts. To facilitate crystallization, we isolated a nanobody (VHH-BS2) from an alpaca
immunized with the TorsinAEQ-LULL1 complex. A stable, heterotrimeric complex of TorsinAEQ-
eLife digest A group of enzymes known as the AAA+ ATPase family have a wide variety of roles
in the cell. They are able to break down a molecule called ATP and use the energy released to
change the structure of other ‘target’ molecules. TorsinA is one such AAA+ ATPase and is found
primarily in nerve cells inside two cell compartments called the endoplasmic reticulum and the
perinuclear space. There, TorsinA interacts with one of two proteins – called LULL1 and LAP1 – that
activate TorsinA. In this respect, TorsinA differs from other members of the AAA+ ATPase family,
which can activate themselves without the need for additional proteins.
TorsinA and other enzymes are made up of building blocks called amino acids. Mutant forms of
TorsinA that have lost a particular amino acid cause primary dystonia, an incurable neuromuscular
disease. This amino acid is needed for TorsinA to interact with LULL1 and LAP1. Previous studies
have revealed the 3D structure of LAP1 on its own, but the structure of TorsinA remained unknown.
One way to study the structure of enzymes is to use a technique called X-ray crystallography. The
first step in this technique is to make crystals of the protein of interest. However, it has proved
difficult to make crystals of TorsinA. Demircioglu et al. have addressed this problem by using X-ray
crystallography to investigate the structure of TorsinA when it is bound to LULL1. The experiments
used a small molecule known as a nanobody that can specifically recognize the human TorsinA
enzyme. The nanobody helped TorsinA to stay attached to LULL1 and form the crystals needed for
X-ray crystallography.
The 3D structures reveal how TorsinA and LULL1 interact in a high level of detail, helping to
explain how TorsinA differs from other AAA+ ATPases. In addition, by comparing how normal
TorsinA and the mutant form interact with LULL1, Demircioglu et al. provide more evidence that
primary dystonia is likely to be caused by the improper activation of TorsinA. The subtle differences
revealed by these structures could be exploited to develop new drugs to fight this disease in the
future.
DOI: 10.7554/eLife.17983.002
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 2 of 14
Research article Biophysics and Structural Biology Cell Biology
Figure 1. Architecture of the TorsinA-LULL1 complex. (A) Schematic diagrams of TorsinA and LULL1. Important residues and sequence motifs are
indicated. The colored areas mark the crystallized segments. Large and small domains of TorsinA are colored in purple and pink, respectively. SS, signal
sequence; H, hydrophobic region; TM, transmembrane helix. (B) Cartoon representation of the TorsinA-LULL1 complex in two orientations. Color-
coding as in (A). A nanobody (VHH-BS2, grey; complementarity determining regions, red) was used as a crystallization chaperone. Numbers refer to
secondary structure elements. (C) Close-up of the ATP binding site. Key residues are labeled. 2FoFc electron density contoured at 2s displayed as
grey mesh. (D) Close-up of the proximal cysteines 280 and 319 next to the adenine base of the bound ATP. 2FoFc electron density is contoured at 1s.
The cysteine pair adopts three alternate conformations, but remains reduced in all of them.
DOI: 10.7554/eLife.17983.003
Figure 1 continued on next page
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 3 of 14
Research article Biophysics and Structural Biology Cell Biology
LULL1-VHH-BS2 was readily crystallized in the presence of ATP. We collected a 1.4 A˚ dataset and
solved the structure by molecular replacement, using the LULL1-homolog LAP1 and a VHH template
as search models (Sosa et al., 2014) (Materials and methods, Table 1). TorsinAEQ adopts a typical
AAA+ ATPase fold (Figure 1B, Figure 1—figure supplement 1). The N-terminal nucleotide-binding
or large domain (residues 55–271) is composed of a central five-stranded, parallel b-sheet sur-
rounded by 8 a-helices. A small three-helix bundle at its C-terminus (residues 272–332), forms critical
contacts with LULL1. An ATP molecule is bound in the manner characteristic of P-loop NTPases
(Wendler et al., 2012). The Walker A and B motifs are positioned to mediate the requisite nucleo-
tide interactions, with sensor 1 and sensor 2 regions sensing the g-phosphate and thus the nucleo-
tide state (Figure 1C). The luminal LULL1 activation domain (residues 236–470) adopts an AAA+-like
conformation, very similar to its paralog LAP1 (rmsd 1.04 A˚ over 211 Ca positions, Figure 1—figure
supplement 1). The AAA+-like domain comprises a central b-sheet embedded within six a-helices
(Figure 1B). A C-terminal small domain is not present. Similar to LAP1, an intramolecular disulfide
bond forms at the C terminus of LULL1, between conserved residues C310 and C468 (Figure 1—fig-
ure supplements 1,3). Characteristically, LULL1 lacks nucleotide binding due to the absence of
Walker A and B motifs (Sosa et al., 2014). LULL1 forms a composite nucleotide-binding site with
TorsinA by providing arginine residue 449 (‘arginine finger’) at the base of helix a5 (Figure 1C). The
arginine finger activates ATP hydrolysis by TorsinA (Brown et al., 2014; Sosa et al., 2014). The small
domain of TorsinA, including helix a7 featuring glutamates 302 and 303, is intimately involved in
LULL1 binding. Nanobody VHH-BS2 binds both TorsinA and LULL1 at a shallow groove (Figure 1B,
Figure 1—figure supplement 4). Nanobodies contain three complementarity determining regions
(CDRs), with CDR3 most often making critical contacts with the antigen (Muyldermans, 2013).
Indeed, the long CDR3 of VHH-BS2 (residues 97–112) is the main binding element in the complex.
AAA+ ATPases are organized into a number of structurally defined clades (Erzberger and
Berger, 2006; Iyer et al., 2004), distinguished by shared structural elements. Comparison with
other AAA+ ATPase structures shows that TorsinA fits best into a clade that also contains the bacte-
rial proteins HslU, ClpA/B, ClpX, and Lon (HCLR clade), all of which are involved in protein degrada-
tion or remodeling (Erzberger and Berger, 2006). These AAA+ family members share a b-hairpin
insertion that precedes the sensor-I region (Figure 1—figure supplement 1). TorsinA also contains
this structural element, but it adopts a distinctly different orientation compared to other members of
the clade; however, the pre-sensor I region may be affected by crystal packing in our structure. Two
other distinct regions are present. The protein degrading or remodeling AAA+ ATPases all form
hexameric rings with a central pore (Hanson and Whiteheart, 2005; Olivares et al., 2016;
White and Lauring, 2007). ‘Pore loops’ in each subunit, conserved elements positioned between
strand b2 and helix a2, are critical for threading the protein substrates through the ring (Sauer and
Baker, 2011). Torsins are devoid of a pore loop consensus motif (Figure 1—figure supplements
2,5). TorsinA has two cysteines (Cys280, and Cys 319, which is part of the sensor-II motif), positioned
near the adenine base of the ATP molecule (Figure 1D). These cysteines do not form a disulfide
bridge in our structure. However, the conservation of Cys280 and the Gly-Cys-Lys sensor-II motif at
position 318–320 (Figure 1—figure supplements 2,5) indicates an important functional role. A
Figure 1 continued
The following figure supplements are available for figure 1:
Figure supplement 1. Structural comparisons.
DOI: 10.7554/eLife.17983.004
Figure supplement 2. Phylogenetic analysis of Torsins.
DOI: 10.7554/eLife.17983.005
Figure supplement 3. Phylogenetic analysis of LAP1/LULL1.
DOI: 10.7554/eLife.17983.006
Figure supplement 4. Nanobody interaction.
DOI: 10.7554/eLife.17983.007
Figure supplement 5. Comparison of sequence motifs of AAA+ ATPases.
DOI: 10.7554/eLife.17983.008
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 4 of 14
Research article Biophysics and Structural Biology Cell Biology
redox activity as part of the ATPase cycle therefore seems highly likely, as has been previously spec-
ulated (Zhu et al., 2008, 2010).
The interaction of TorsinA with its ATPase activators LULL1 and LAP1 is of particular importance,
as a prominent mutation causing primary dystonia–the deletion of glutamate 302 or 303–weakens
these interactions (Naismith et al., 2009; Zhao et al., 2013; Zhu et al., 2010). But why and how?
The TorsinA-LULL1 interface extends over an area of 1439 A˚2. The main structural elements involved
in this interaction are the nucleotide-binding region as well as the small domain of TorsinA, and heli-
ces a0, a2, a4 and a5 of LULL1 (Figure 1, Figure 1—figure supplements 2,3, Figure 2A). The exact
position of the small domain of TorsinA relative to the large domain is likely dictated by the sensor II
motif, preceding a8, which directly contacts the g-phosphate of ATP through Lys 320, thus serving
as an anchor point. A switch to ADP presumably weakens this connection, such that the small
domain would become more loosely attached to the large domain. This could explain the observed
ATP-dependency of LAP1/LULL1 binding (Goodchild and Dauer, 2005; Naismith et al., 2009;
Table 1. X-ray data collection and refinement statistics.
TorsinA-LULL1233-470 TorsinADE-LULL1233-470
PDB Code 5J1S 5J1T
Data collection
Space group P212121 P212121
Cell dimensions
a, b, c (A˚) 75.7, 90.7, 105.1 75.4, 88.4, 105.3
a, b, g (˚) 90.0, 90.0, 90.0 90.0, 90.0, 90.0
Resolution (A˚) 61–1.40 (1.45–1.40)* 68–1.40 (1.45–1.40)
Rsym 0.06 (0.88) 0.10 (1.98)
Rpim 0.03 (0.43) 0.03 (0.60)
I / s 33.0 (1.5) 30.8 (1.3)
Completeness (%) 94.7 (67.5) 97.9 (96.5)
Redundancy 5.7 (4.4) 12.4 (11.3)
CC(1/2) 1.00 (0.65) 1.00 (0.58)
Refinement
Resolution (A˚) 61.4–1.40 67.7–1.40
No. reflections 132956 134333
Rwork / Rfree 0.143/0.188 0.148/0.177
No. atoms 5898 5927
Protein 5241 5244
Ligand/ion 35 47
Water 622 636
B factors (A˚2)
Protein 31.3 24.0
Ligand/ion 23.2 17.2
Water 43.1 33.6
r.m.s. deviations
Bond lengths (A˚) 0.014 0.017
Bond angles (˚) 1.25 1.71
Ramachandran
Favored/allowed/outliers (%) 98.0/1.7/0.0 98.6/1.4/0.0
*Values in parentheses are for highest-resolution shell. One crystal was used for each dataset.
DOI: 10.7554/eLife.17983.009
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 5 of 14
Research article Biophysics and Structural Biology Cell Biology
Zhao et al., 2013; Zhu et al., 2010). Within the small domain, helix a7, the following loop, and the
terminal helix a8 contain all the critical residues. Glutamate 302 and 303 are positioned at the very
end of helix a7, and both are involved in TorsinA contacts. Specifically, Glu 303 forms a prominent
charge interaction with Arg 276 of LULL1. TorsinA Lys113 – LULL1 Glu385, TorsinA Asp316 - LULL1
Arg419, TorsinA Lys317 - LULL1 Glu415, TorsinA Asp327 - LULL1 Lys283 are additional charge
interactions.
To investigate the atomic details of the weakened binding of TorsinADE to LAP1/LULL1, and thus
the molecular basis of primary dystonia, we made use of the observation that VHH-BS2 also stabil-
izes the TorsinAEQDE(ATP)-LULL1 interaction. We were able to crystallize TorsinAEQDE(ATP)-LULL1-
VHH-BS2 and determine its structure at a resolution of 1.4 A˚. Not surprisingly, the overall structure
is almost identical to the wild-type protein (0.34 A˚ rmsd over 276 Ca atoms for TorsinA, 0.27 A˚
rmsd over 226 Ca atoms for LULL1), except for critical differences in the TorsinA-LULL1 interface
(Figure 2A). The principal difference is that helix a7 is shortened due to the missing Glu 303, with a
slight–but significant–restructuring of the loop that follows to establish the connection with helix a8.
For future reference, we suggest renaming the DE mutation DE303, rather than DE302/303, since the
position of Glu 302 is effectively unchanged. In the dystonia mutant, the TorsinA Glu 303 – LULL1
Arg 276 charge interaction is lost, and the hydrogen-bonding network involving TorsinA Glu 302,
Phe 306 and Arg312, as well as LULL1 Arg412 and Glu416 is disrupted (Figure 2A). To determine
the importance of different TorsinA residues for LULL1 binding, we performed a co-purification assay
Figure 2. Analysis of the TorsinA-LULL1 interface. (A) Side-by-side comparison of TorsinA-ATP-LULL1 (left) and
TorsinADE-ATP-LULL1 (right). Zoomed insets show the atomic details of the interactions between TorsinA/
TorsinADE and LULL1, with a focus on the DE303 area. (B and C) Mutational analysis of the TorsinA-LULL1
interface. Substitution or deletion of residues involved in TorsinA-LULL1 binding were probed using a Ni-affinity
co-purification assay with recombinant, bacterial-expressed protein. Only TorsinA is His-tagged. SDS-PAGE
analysis is shown. Lack of binding is observed by the absence of complex (uncomplexed His-tagged TorsinA is
insoluble). t, total lysate, e, Ni eluate. Asterisk denotes an unrelated contaminant.
DOI: 10.7554/eLife.17983.010
The following figure supplement is available for figure 2:
Figure supplement 1. Structural mapping of mutations causing dystonia.
DOI: 10.7554/eLife.17983.011
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 6 of 14
Research article Biophysics and Structural Biology Cell Biology
(Figure 2B,C). His-tagged, ATP-trapped TorsinAEQ (residues 51–332) and mutants thereof were
recombinantly co-expressed with LULL1 (residues 233–470), but without VHH-BS2, in bacteria. Bind-
ing was tested in a co-purification assay using Ni-affinity. The TorsinEQDE303 mutation abolishes
binding in this assay, as expected (Figure 2B). Since unbound TorsinAEQ is largely insoluble, absence
of binding is not registered as an appearance of TorsinAEQ alone, but rather as a lack of eluted pro-
tein complex altogether. Eliminating the salt bridge between TorsinA Glu303 and LULL1 Arg276
does not disrupt the TorsinA-LULL1 interaction (Figure 2B). However, DMet304 and DThr305 both
phenocopy DE303 in abolishing LULL1 binding (Figure 2C). This is in full agreement with published
in vivo data using similar mutants (Goodchild and Dauer, 2004). The intricate network of interac-
tions of the a7-a8 loop of TorsinA is crucial for LULL1 binding. Since the DE mutation causes a local
change within the small domain of TorsinA rather than protein misfolding, it may be possible to res-
cue binding by developing a small molecule that resurrects the weakened TorsinADE-LAP1/LULL1
interaction.
Although TorsinADE303 is the most prevalent mutation that causes primary dystonia, it is not the
only one (Laudermilch and Schlieker, 2016; Rose et al., 2015). We examined the structural conse-
quence of all known mutations (Figure 2—figure supplement 1, Table 2). Based on our structural
data, we strongly predict that most mutations likely cause protein misfolding or they weaken or
abolish LAP1/LULL1 binding. Conversely, the two dystonia-mutations found in LAP1 presumably
affect torsin interaction. Our structural data, therefore, clearly support the hypothesis that improper
torsin activation is the likely cause of primary dystonia (Kim et al., 2010).
Discussion
The biological function of TorsinA remains enigmatic (Granata et al., 2011; Jokhi et al., 2013;
Liang et al., 2014; Nery et al., 2008, 2011). Because TorsinA belongs to the AAA+ ATPase super-
family, with specific homology to the bacterial proteins HslU, ClpX, ClpA/B and Lon, it is generally
assumed that TorsinA is involved in protein remodeling or protein degradation (Laudermilch and
Schlieker, 2016; Rose et al., 2015). However, a substrate of TorsinA has yet to be identified.
The TorsinA structure enables a more thorough comparison to other AAA+ ATPases, particularly
with regard to the functionally relevant oligomerization state. After the discovery that LAP1/LULL1
are Arg-finger containing TorsinA activators with a AAA+-like structure, it seemed reasonable to
suggest that TorsinA and LAP1/LULL1 likely form heterohexameric rings ((TorsinA-ATP-LAP1/
LULL1)3) in order to function (Brown et al., 2014; Sosa et al., 2014). However, the predominant
oligomeric form of recombinant TorsinA-ATP-LAP1/LULL1 complex in vitro and in solution is the het-
erodimer (Brown et al., 2014; Sosa et al., 2014). In addition, torsin variants have been reported to
occur in various oligomeric forms as detected by Blue Native PAGE (BN-PAGE) (Goodchild et al.,
Table 2. Dystonia mutations.
Protein Mutation Structural consequence Reference
TorsinA DE302/303 Weakened LAP1/LULL1 binding (Ozelius et al., 1997)
TorsinA DF323-Y328 Weakened LAP1/LULL1 binding (Leung et al., 2001)
TorsinA R288Q Weakened LAP1/LULL1 binding (Zirn et al., 2008)
TorsinA F205I Folding problem (Calakos et al., 2010)
TorsinA D194V Change to the conserved, noncatalytic interface (Cheng et al., 2014)
TorsinA DA14-P15 Improper cellular targeting (Vulinovic et al., 2014)
TorsinA E121K Charge inversion at the membrane proximal interface (Vulinovic et al., 2014)
TorsinA V129I Folding problem (Dobricˇic´ et al., 2015)
TorsinA D216H (modifier) Surface change; consequence unclear (Kamm et al., 2008; Kock et al., 2006)
LAP1 c.186deiG (p.E62fsTer25) Lack of the luminal activation domain of LAP1 (Kayman-Kurekci et al., 2014)
LAP1 E482A* Improper folding (Dorboz et al., 2014)
*Assesment based on the equivalent residue in LULL1 (E368).
DOI: 10.7554/eLife.17983.012
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 7 of 14
Research article Biophysics and Structural Biology Cell Biology
2015; Jungwirth et al., 2010; Vander Heyden et al., 2009). Our structure now raises doubts about
the physiological relevance of a heterohexameric ring (Figure 3). First, we note that the small
domain of TorsinA is essential for LAP1/LULL1 binding (Figure 2C). This is reminiscent of the related
HCLR AAA+ clade members where the small domain is known to be critical for hexamerization
(Bochtler et al., 2000; Mogk et al., 2003). The importance of the small domain for oligomerization
in the context of torsins has also been discussed recently (Rose et al., 2015). Neither LAP1 nor
LULL1 harbor a small domain, arguing against formation of a stable heteromeric ring, or, alterna-
tively, suggesting a ring of substantially different architecture. Second, ring formation is important
for AAA+ ATPases that thread their protein substrate through a central pore for refolding or for
degradation. This central pore is lined with conserved ‘pore loops’ that are essential for function
(White and Lauring, 2007). Neither TorsinA and its homologs, nor LAP1/LULL1 have ‘pore loop’
equivalents (Figure 1—figure supplement 5). TorsinA is therefore unlikely to actually employ a pep-
tide threading mechanism that involves a central pore. Third, the surface conservation of LAP1/
LULL1 also argues against a heteromeric ring assembly. Although the catalytic, ATP-containing inter-
face with TorsinA is well-conserved, the presumptive non-catalytic, nucleotide-free interface is not
(Figure 3B). Importantly and in contrast to LAP1/LULL1, the same analysis for TorsinA shows that its
‘backside’ is conserved. TorsinA may therefore interact in homotypic fashion with TorsinA, with other
torsin homologs, or even with an additional, yet unidentified protein. This could mean that the
Figure 3. Oligomerization of TorsinA-LULL1. (A) Left, Schematic representation of a hypothetical heterohexameric (TorsinA-LULL1)3 ring model, in
analogy to canonical AAA+ ATPases. White star represents ATP. Since LULL1 cannot bind a nucleotide, there would be three catalytic (nucleotide-
bound) and three non-catalytic interfaces per ring. Open-book representation of the catalytic interface between TorsinA and LULL1, as seen in this
study. Black line marks the outline of the interface. Color gradient marks conservation across diverse eukaryotes. (B) The same analysis as in (A), but for
the hypothetical ‘non-catalytic’ interface. The interface model on the right is based on swapping the TorsinA and LULL1 positions in the TorsinA-LULL1
complex.
DOI: 10.7554/eLife.17983.013
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 8 of 14
Research article Biophysics and Structural Biology Cell Biology
previously observed hexameric assemblies (Goodchild et al., 2015; Jungwirth et al., 2010;
Sosa et al., 2014) may only contain one LAP1/LULL1 unit, and multiple torsin units, a property that
the employed assays would not have differentiated. It is also possible, that the reported hexameric
assemblies reflect a vestigial, yet physiologically irrelevant property, perhaps just of the evolutionary
origin of the Torsin-LAP1/LULL1 system. Taking all the existing data into account, it is suggestive
that TorsinA may be an exceptional AAA+ ATPase in that it simply acts as a heterodimer, together
with LAP1 or LULL1 functioning as an activator. As long as the biological function and the substrate
for TorsinA are unclear, however, the physiologically relevant oligomeric state of TorsinA ultimately
remains a matter of speculation. Given the unique properties of TorsinA, keeping an open mind
about TorsinA assembly into its functional state is called for, as it may well differ more than antici-
pated from well-studied AAA+ ATPase systems.
The observation that the nanobody VHH-BS2 stabilizes the TorsinADE303-LULL1 suggests that it
could possibly be used directly as a therapeutic. After all, it could directly rescue TorsinA activity.
There are, however, at least two major problems. First, VHH-BS2 only recognizes the TorsinA- (or
TorsinADE303-) LULL1 complex, but not the homologous TorsinA-LAP1 complex. The function of
LULL1 is still poorly understood, but a knockdown does not generate an NE blebbing phenotype
(Goodchild et al., 2015; Turner et al., 2015; Vander Heyden et al., 2009), which is symptomatic
for a TorsinA knockout (Goodchild et al., 2005) or a LAP1 knockdown (Kim et al., 2010). Therefore,
resurrecting activation of TorsinADE303 via LULL1 is unlikely to ameliorate the dystonia phenotype.
Furthermore, the nanobody interaction site on the TorsinA-LULL1 interface is very likely oriented
toward the ER membrane, which can be inferred from the relative positions of the membrane anchor
of LULL1 and the hydrophobic, likely membrane-proximal N-terminal region of TorsinA. These topo-
logical restraints suggest that the nanobody will not bind in vivo, but that it is of significant use for in
vitro studies.
Materials and methods
Constructs, protein expression and purification
DNA sequences encoding human TorsinA (residues 51–332) and the luminal domain of human
LULL1 (residues 233–470) were cloned into a modified ampicillin resistant pETDuet-1 vector (EMD
Millipore). TorsinA, N-terminally fused with a human rhinovirus 3C protease cleavable 10xHis-7xArg
tag, was inserted into the first multiple cloning site (MCS), whereas the untagged LULL1 was inserted
into the second MCS. Mutations on TorsinA and LULL1 were introduced by site-directed mutagene-
sis. The untagged VHH-BS2 nanobody was cloned into a separate, modified kanamycin resistant
pET-30b(+) vector (EMD Biosciences).
To co-express TorsinA (EQ or EQ/DE), LULL1 and VHH-BS2 for crystallization, the E. coli strain
LOBSTR(DE3) RIL (Kerafast, Boston MA) (Andersen et al., 2013) was co-transformed with the two
constructs described above. Cells were grown at 37˚C in lysogeny broth (LB) medium supplemented
with 100 mg ml1 ampicillin, 25mg ml1 kanamaycin and 34 mg ml1 chloramphenicol until an optical
density (OD600) of 0.6–0.8 was reached, shifted to 18˚C for 20 min, and induced overnight at 18˚C
with 0.2 mM isopropyl b-D-1-thiogalactopyranoside (IPTG). The bacterial cultures were harvested by
centrifugation, suspended in lysis buffer (50 mM HEPES/NaOH pH 8.0, 400 mM NaCl, 40 mM imid-
azole, 10 mM MgCl2, and 1 mM ATP) and lysed with a cell disruptor (Constant Systems). The lysate
was immediately mixed with 0.1 M phenylmethanesulfonyl fluoride (PMSF) (50 ml per 10 ml lysate)
and 250 units of TurboNuclease (Eton Bioscience), and cleared by centrifugation. The soluble frac-
tion was gently mixed with Ni-Sepharose 6 Fast Flow (GE Healthcare) resin for 30 min at 4˚C. After
washing with the lysis buffer, bound protein was eluted in elution buffer (10 mM HEPES/NaOH pH
8.0, 150 mM NaCl, 300 mM imidazole, 10 mM MgCl2, and 1 mM ATP). The eluted protein complex
was immediately purified by size exclusion chromatography on a Superdex S200 column (GE Health-
care) equilibrated in running buffer (10 mM HEPES/NaOH pH 8.0, 150 mM NaCl, 10 mM MgCl2,
and 0.5 mM ATP). Following the tag removal by 10xHis-7xArg-3C protease, the fusion tags and the
protease were separated from the complex by cation-exchange chromatography on a HiTrapS col-
umn (GE Healthcare) using a linear NaCl gradient. The flow-through from the cation-exchange chro-
matography, containing the protein complex, was purified again by size exclusion chromatography
on a Superdex S200 column as at the previous step.
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 9 of 14
Research article Biophysics and Structural Biology Cell Biology
For the non-structural analysis of TorsinA and LULL1 variants, the pETDuet-1-based expression
plasmid was transformed into LOBSTR(DE3) RIL cells without co-expressing nanobody VHH-BS2.
Ni2+-affinity purification was performed as described above and bound protein was eluted. Aliquots
from the Ni2+-eluate and the total lysate were collected and analyzed by SDS-PAGE gel
electrophoresis.
Crystallization
Purified TorsinAEQ-LULL1-VHH-BS2 and TorsinAEQDE-LULL1-VHH-BS2 complexes were concentrated
up to 4–4.5 mg/ml and supplemented with 2 mM ATP prior to crystallization. The TorsinAEQ contain-
ing complex crystallized in 13% (w/v) polyethylene glycol (PEG) 6000, 5% (v/v) 2-Methyl-2,4-pentane-
diol, and 0.1 M MES pH 6.5. The TorsinAEQDE containing complex crystallized in 19% (w/v) PEG
3350, 0.2 M AmSO4, and 0.1 M Bis-Tris-HCl pH 6.5. Crystals of both complexes grew at 18˚C in
hanging drops containing 1 ml of protein and 1 ml of mother liquor. Clusters of diffraction quality,
rod-shaped crystals formed within 3–5 days. Single crystals were briefly soaked in mother liquor sup-
plemented with 20% (v/v) glycerol for cryoprotection and flash-frozen in liquid nitrogen.
Data collection and structure determination
X-ray data were collected at NE-CAT beamline 24-ID-C at Argonne National Laboratory. Data reduc-
tion was performed with the HKL2000 package (Otwinowski and Minor, 1997), and all subsequent
data-processing steps were carried out using programs provided through SBGrid (Morin et al.,
2013). The structure of the TorsinAEQ-LULL1-VHH-BS2 complex was solved by molecular replace-
ment (MR) using the Phaser-MR tool from the PHENIX suite (Adams et al., 2010). A three-part MR
solution was easily obtained using a sequential search for models of LULL1, VHH-BS2, and TorsinA.
The LULL1 model was generated based on the published human LAP1 structure (PDB 4TVS, chain
A), using the Sculptor utility of the PHENIX suite (LULL1241–470 and LAP1356–583 share 64% sequence
identity). The VHH-BS2 model was based on VHH-BS1 (PDB 4TVS, chain a) after removing the com-
plementarity determining regions (CDRs). The poly-Ala model of TorsinA was generated based on E.
coli ClpA (PDB 1R6B) using the MODELLER tool of the HHpred server (So¨ding et al., 2005). The
asymmetric unit contains one TorsinAEQ-LULL1-VHH-BS2 complex. Iterative model building and
refinement steps gradually improved the electron density maps and the model statistics. The stereo-
chemical quality of the final model was validated by Molprobity (Chen et al., 2010). TorsinAEQDE-
LULL1-VHH-BS2 crystallized in the same unit cell. Model building was carried starting from a trun-
cated TorsinAEQ-LULL1-VHH-BS2 structure. All manual model building steps were carried out with
Coot (Emsley et al., 2010), and phenix.refine was used for iterative refinement. Two alternate con-
formations of a loop in LULL1 (residues 428–438) were detected in the FoFc difference electron
density maps of both structures, and they were partially built. For comparison, the cysteine residues
of TorsinA at the catalytic site (residues 280 and 319 in the TorsinAEQ structure) were built in the
reduced and the oxidized states, respectively. Building them as oxidized, disulfide-bridged residues
consistently produced substantial residual FoFc difference density, which disappeared assuming a
reduced state. Statistical parameters of data collection and refinement are all given in Table 1.
Structure figures were created in PyMOL (Schro¨dinger LLC).
Bioinformatic analysis
Torsin and LAP1/LULL1 sequences were obtained via PSI-BLAST (Altschul et al., 1997) and Back-
phyre searches (Kelley and Sternberg, 2009). Transmembrane domains were predicted using the
HMMTOP tool (Tusna´dy and Simon, 2001). LAP1/LULL1 proteins were distinguished based on the
calculated isoelectric point (pI) of their extra-luminal portions. The intranuclear domain of LAP1 has a
characteristically high pI of ~8.5–10 due to a clustering of basic residues, while the cytoplasmic
domain of LULL1 is distinctively more acidic. Multiple sequence alignments were performed using
MUSCLE (Edgar, 2004), and visualized by Jalview (Waterhouse et al., 2009). To illustrate evolution-
ary conservation on TorsinA and LULL1 surfaces, conservation scores for each residue were calcu-
lated using the ConSurf server with default parameters (Glaser et al., 2003).
The sequences, which were used to generate the multiple sequence alignments, were also used
for preparing the sequence logos of Torsins and LAP1/LULL1 in Figure 1—figure supplement 5. To
obtain the sequence logo of the HCLR clade AAA+ ATPases, Escherichia coli ClpA-D2 (residues
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 10 of 14
Research article Biophysics and Structural Biology Cell Biology
458–758), Escherichia coli ClpB-D2 (residues 568–857), Bacillus subtilis ClpE-D2 (residues 409–699),
Saccharomyces cerevisiae Hsp104-D2 (residues 578–868), Escherichia coli HslU (residues 13–443),
Bacillus subtilis HslU (residues 15–455), Streptomyces coelicolor ClpX (residues 71–409), Drosophila
melanogaster ClpX (residues 199–634), Escherichia coli Lon (residues 320–580), Caenorhabditis ele-
gans Lon (residues 476–771), Thermus thermophilus ClpB-D2 (residues 536–845), Escherichia coli
ClpX (residues 64–403), Helicobacter pylori ClpX (residues 77–430), Haemophilus influenza HslU (1–
444), Bacillus subtilis Lon (residues 300–590), Bacillus subtilis ClpC-D2 (residues 486–802), Saccharo-
myces cerevisiae Hsp78-D2 (residues 482–794) and Arabidopsis thaliana Hsp101-D2 (residues 547–
849) sequences were used. All sequence logos were generated using WebLogo (Crooks et al.,
2004).
Generation and selection of nanobodies
The purified human TorsinAEQ-LULL1 complex was injected into a male alpaca (Lama pacos) for
immunization. Generation and screening of nanobodies was carried out as previously described
(Sosa et al., 2014). Each of the selected nanobodies was subcloned into a pET-30b(+) vector with a
C-terminal His6-tag. Each nanobody was bacterially expressed and Ni
2+-affinity purified essentially as
described (see above). Different from the TorsinA-containing preparations, MgCl2 and ATP were
eliminated from all buffer solutions. The Ni2+-eluate was purified via size exclusion chromatography
on a Superdex S75 column (GE Healthcare) in running buffer (10 mM HEPES/NaOH pH 8.0, 150 mM
NaCl). Nanobody binding was validated by size exclusion chromatography on a 10/300 Superdex
S200 column in 10 mM HEPES/NaOH pH 8.0, 150 mM NaCl, 10 mM MgCl2 and 0.5 mM ATP. Equi-
molar amounts of TorsinAEQ-LULL1 and TorsinAEQ-LULL1-VHH were loaded and nanobody binding
was monitored by a shift in the elution profile and via SDS-PAGE analysis. After validating VHH-BS2
interaction with TorsinAEQ-LULL1, the C-terminal His6-tag of VHH-BS2 was removed from the pET-
30b(+) vector for co-purification experiments.
Acknowledgements
The authors thank Ulrike Kutay for many helpful discussions pertinent to this study. The work in the
Schwartz lab was supported by the Foundation for Dystonia Research and a National Institutes of
Health grant (AR065484). Work in the lab of HLP was supported by an NIH Director’s Pioneer award.
Structural data are based upon research conducted at the Northeastern Collaborative Access Team
beamlines, which are funded by the National Institute of General Medical Sciences from the National
Institutes of Health (P41 GM103403). This research used resources of the Advanced Photon Source,
a US Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Sci-
ence by Argonne National Laboratory under Contract No. DE-AC02-06CH11357.
Additional information
Competing interests
FED, BAS and TUS: Filed a provisional patent application protecting the use of the crystal structures
(U.S.P.T.O. No. 62/330,683). The other authors declare that no competing interests exist.
Funding
Funder Grant reference number Author
National Institutes of Health AR065484 Hidde L Ploegh
Thomas U Schwartz
National Institutes of Health Director’s Pioneer award Hidde L Ploegh
Foundation for Dystonia Re-
search
Thomas U Schwartz
National Institute of General
Medical Sciences
P41 GM103403 Thomas U Schwartz
The funders had no role in study design, data collection and interpretation, or the decision to
submit the work for publication.
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 11 of 14
Research article Biophysics and Structural Biology Cell Biology
Author contributions
FED, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or
revising the article; BAS, Acquisition of data, Analysis and interpretation of data; JI, Acquisition of
data, Drafting or revising the article; HLP, Analysis and interpretation of data, Drafting or revising
the article; TUS, Conception and design, Analysis and interpretation of data, Drafting or revising the
article
Author ORCIDs
F Esra Demircioglu, http://orcid.org/0000-0002-3866-2742
Thomas U Schwartz, http://orcid.org/0000-0001-8012-1512
Additional files
Major datasets
The following datasets were generated:
Author(s) Year Dataset title Dataset URL
Database, license,
and accessibility
information
F Esra Demircioglu,
Thomas U Schwartz
2016 TorsinA-LULL1 complex, H. sapiens,
bound to VHH-BS2
http://www.rcsb.org/
pdb/explore/explore.do?
structureId=5J1S
Publicly available at
RCSB Protein Data
Bank (accession no.
5J1S)
F Esra Demircioglu,
Thomas U Schwartz
2016 TorsinAdeltaE-LULL1 complex, H.
sapiens, bound to VHH-BS2
http://www.rcsb.org/
pdb/explore/explore.do?
structureId=5J1T
Publicly available at
RCSB Protein Data
Bank (accession no.
5J1T)
References
Adams PD, Afonine PV, Bunko´czi G, Chen VB, Davis IW, Echols N, Headd JJ, Hung LW, Kapral GJ, Grosse-
Kunstleve RW, McCoy AJ, Moriarty NW, Oeffner R, Read RJ, Richardson DC, Richardson JS, Terwilliger TC,
Zwart PH. 2010. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta
Crystallographica Section D Biological Crystallography 66:213–221. doi: 10.1107/S0907444909052925
Altschul SF, Madden TL, Scha¨ffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ. 1997. Gapped BLAST and PSI-
BLAST: a new generation of protein database search programs. Nucleic Acids Research 25:3389–3402. doi: 10.
1093/nar/25.17.3389
Andersen KR, Leksa NC, Schwartz TU. 2013. Optimized E. coli expression strain LOBSTR eliminates common
contaminants from His-tag purification. Proteins 81:1857–1861. doi: 10.1002/prot.24364
Bochtler M, Hartmann C, Song HK, Bourenkov GP, Bartunik HD, Huber R. 2000. The structures of HsIU and the
ATP-dependent protease HsIU-HsIV. Nature 403:800–805. doi: 10.1038/35001629
Breakefield XO, Blood AJ, Li Y, Hallett M, Hanson PI, Standaert DG, Li AJ, Hanson M. 2008. The
pathophysiological basis of dystonias. Nature Reviews Neuroscience 9:222–234. doi: 10.1038/nrn2337
Brown RSH, Zhao C, Chase AR, Wang J, Schlieker C. 2014. The mechanism of Torsin ATPase activation. PNAS
111:E4822–E4831. doi: 10.1073/pnas.1415271111
Calakos N, Patel VD, Gottron M, Wang G, Tran-Viet KN, Brewington D, Beyer JL, Steffens DC, Krishnan RR,
Zu¨chner S. 2010. Functional evidence implicating a novel TOR1A mutation in idiopathic, late-onset focal
dystonia. Journal of Medical Genetics 47:646–650. doi: 10.1136/jmg.2009.072082
Chen VB, Arendall WB, Headd JJ, Keedy DA, Immormino RM, Kapral GJ, Murray LW, Richardson JS, Richardson
DC. 2010. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallographica
Section D Biological Crystallography 66:12–21. doi: 10.1107/S0907444909042073
Cheng FB, Feng JC, Ma LY, Miao J, Ott T, Wan XH, Grundmann K. 2014. Combined occurrence of a novel
TOR1A and a THAP1 mutation in primary dystonia. Movement Disorders 29:1079–1083. doi: 10.1002/mds.
25921
Crooks GE, Hon G, Chandonia JM, Brenner SE. 2004. WebLogo: a sequence logo generator. Genome Research
14:1188–1190. doi: 10.1101/gr.849004
Dobricˇic´ V, Kresojevic´ N, Zˇarkovic´ M, Tomic´ A, Marjanovic´ A, Westenberger A, Cvetkovic´ D, Svetel M, Novakovic´
I, Kostic´ VS. 2015. Phenotype of non-c.907_909delGAG mutations in TOR1A: DYT1 dystonia revisited.
Parkinsonism & Related Disorders 21:1256–1259. doi: 10.1016/j.parkreldis.2015.08.001
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 12 of 14
Research article Biophysics and Structural Biology Cell Biology
Dorboz I, Coutelier M, Bertrand AT, Caberg JH, Elmaleh-Berge`s M, Laine´ J, Stevanin G, Bonne G, Boespflug-
Tanguy O, Servais L. 2014. Severe dystonia, cerebellar atrophy, and cardiomyopathy likely caused by a
missense mutation in TOR1AIP1. Orphanet Journal of Rare Diseases 9. doi: 10.1186/s13023-014-0174-9
Edgar RC. 2004. MUSCLE: a multiple sequence alignment method with reduced time and space complexity.
BMC Bioinformatics 5. doi: 10.1186/1471-2105-5-113
Emsley P, Lohkamp B, Scott WG, Cowtan K. 2010. Features and development of Coot. Acta Crystallographica
Section D Biological Crystallography 66:486–501. doi: 10.1107/S0907444910007493
Erzberger JP, Berger JM. 2006. Evolutionary relationships and structural mechanisms of AAA+ proteins. Annual
Review of Biophysics and Biomolecular Structure 35:93–114. doi: 10.1146/annurev.biophys.35.040405.101933
Glaser F, Pupko T, Paz I, Bell RE, Bechor-Shental D, Martz E, Ben-Tal N. 2003. ConSurf: identification of
functional regions in proteins by surface-mapping of phylogenetic information. Bioinformatics 19:163–164. doi:
10.1093/bioinformatics/19.1.163
Goodchild RE, Buchwalter AL, Naismith TV, Holbrook K, Billion K, Dauer WT, Liang CC, Dear ML, Hanson PI.
2015. Access of torsinA to the inner nuclear membrane is activity dependent and regulated in the endoplasmic
reticulum. Journal of Cell Science 128:2854–2865. doi: 10.1242/jcs.167452
Goodchild RE, Dauer WT. 2004. Mislocalization to the nuclear envelope: An effect of the dystonia-causing
torsinA mutation. PNAS 101:847–852. doi: 10.1073/pnas.0304375101
Goodchild RE, Dauer WT. 2005. The AAA+ protein torsinA interacts with a conserved domain present in LAP1
and a novel ER protein. Journal of Cell Biology 168:855–862. doi: 10.1083/jcb.200411026
Goodchild RE, Kim CE, Dauer WT. 2005. Loss of the dystonia-associated protein torsinA selectively disrupts the
neuronal nuclear envelope. Neuron 48:923–932. doi: 10.1016/j.neuron.2005.11.010
Granata A, Koo SJ, Haucke V, Schiavo G, Warner TT. 2011. CSN complex controls the stability of selected
synaptic proteins via a torsinA-dependent process. The EMBO Journal 30:181–193. doi: 10.1038/emboj.2010.
285
Granata A, Warner TT. 2010. The role of torsinA in dystonia. European Journal of Neurology 17 Suppl 1:81–87.
doi: 10.1111/j.1468-1331.2010.03057.x
Hanson PI, Whiteheart SW. 2005. AAA+ proteins: have engine, will work. Nature Reviews Molecular Cell Biology
6:519–529. doi: 10.1038/nrm1684
Iyer LM, Leipe DD, Koonin EV, Aravind L. 2004. Evolutionary history and higher order classification of AAA+
ATPases. Journal of Structural Biology 146:11–31. doi: 10.1016/j.jsb.2003.10.010
Jokhi V, Ashley J, Nunnari J, Noma A, Ito N, Wakabayashi-Ito N, Moore MJ, Budnik V. 2013. Torsin mediates
primary envelopment of large ribonucleoprotein granules at the nuclear envelope. Cell Reports 3:988–995. doi:
10.1016/j.celrep.2013.03.015
Jungwirth M, Dear ML, Brown P, Holbrook K, Goodchild R. 2010. Relative tissue expression of homologous
torsinB correlates with the neuronal specific importance of DYT1 dystonia-associated torsinA. Human Molecular
Genetics 19:888–900. doi: 10.1093/hmg/ddp557
Kamm C, Fischer H, Garavaglia B, Kullmann S, Sharma M, Schrader C, Grundmann K, Klein C, Borggraefe I,
Lobsien E, Kupsch A, Nardocci N, Gasser T. 2008. Susceptibility to DYT1 dystonia in European patients is
modified by the D216H polymorphism. Neurology 70:2261–2262. doi: 10.1212/01.wnl.0000313838.05734.8a
Kayman-Kurekci G, Talim B, Korkusuz P, Sayar N, Sarioglu T, Oncel I, Sharafi P, Gundesli H, Balci-Hayta B, Purali
N, Serdaroglu-Oflazer P, Topaloglu H, Dincer P. 2014. Mutation in TOR1AIP1 encoding LAP1B in a form of
muscular dystrophy: a novel gene related to nuclear envelopathies. Neuromuscular Disorders 24:624–633. doi:
10.1016/j.nmd.2014.04.007
Kelley LA, Sternberg MJ. 2009. Protein structure prediction on the Web: a case study using the Phyre server.
Nature Protocols 4:363–371. doi: 10.1038/nprot.2009.2
Kim CE, Perez A, Perkins G, Ellisman MH, Dauer WT. 2010. A molecular mechanism underlying the neural-
specific mice. PNAS 107:9861–9866. doi: 10.1073/pnas.0912877107
Kock N, Naismith TV, Boston HE, Ozelius LJ, Corey DP, Breakefield XO, Hanson PI, Ozelius HE, Corey LJ. 2006.
Effects of genetic variations in the dystonia protein torsinA: identification of polymorphism at residue 216 as
protein modifier. Human Molecular Genetics 15:1355–1364. doi: 10.1093/hmg/ddl055
Laudermilch E, Schlieker C. 2016. Torsin ATPases: structural insights and functional perspectives. Current
Opinion in Cell Biology 40:1–7. doi: 10.1016/j.ceb.2016.01.001
Leung JC, Klein C, Friedman J, Vieregge P, Jacobs H, Doheny D, Kamm C, DeLeon D, Pramstaller PP, Penney
JB, Eisengart M, Jankovic J, Gasser T, Bressman SB, Corey DP, Kramer P, Brin MF, Ozelius LJ, Breakefield XO.
2001. Novel mutation in the TOR1A (DYT1) gene in atypical early onset dystonia and polymorphisms in
dystonia and early onset parkinsonism. Neurogenetics 3:133–143. doi: 10.1007/s100480100111
Liang CC, Tanabe LM, Jou S, Chi F, Dauer WT. 2014. TorsinA hypofunction causes abnormal twisting movements
and sensorimotor circuit neurodegeneration. Journal of Clinical Investigation 124:3080–3092. doi: 10.1172/
JCI72830
McCullough J, Sundquist WI. 2014. Putting a finger in the ring. Nature Structural & Molecular Biology 21:1025–
1027. doi: 10.1038/nsmb.2928
Mogk A, Schlieker C, Strub C, Rist W, Weibezahn J, Bukau B. 2003. Roles of individual domains and conserved
motifs of the AAA+ chaperone ClpB in oligomerization, ATP hydrolysis, and chaperone activity. Journal of
Biological Chemistry 278:17615–17624. doi: 10.1074/jbc.M209686200
Morin A, Eisenbraun B, Key J, Sanschagrin PC, Timony MA, Ottaviano M, Sliz P. 2013. Collaboration gets the
most out of software. eLife 2:e01456. doi: 10.7554/eLife.01456
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 13 of 14
Research article Biophysics and Structural Biology Cell Biology
Muyldermans S. 2013. Nanobodies: natural single-domain antibodies. Annual Review of Biochemistry 82:775–
797. doi: 10.1146/annurev-biochem-063011-092449
Naismith TV, Dalal S, Hanson PI. 2009. Interaction of torsinA with its major binding partners is impaired by the
dystonia-associated DeltaGAG deletion. Journal of Biological Chemistry 284:27866–27874. doi: 10.1074/jbc.
M109.020164
Nery FC, Armata IA, Farley JE, Cho JA, Yaqub U, Chen P, da Hora CC, Wang Q, Tagaya M, Klein C, Tannous B,
Caldwell KA, Caldwell GA, Lencer WI, Ye Y, Breakefield XO. 2011. TorsinA participates in endoplasmic
reticulum-associated degradation. Nature Communications 2:393 . doi: 10.1038/ncomms1383
Nery FC, Zeng J, Niland BP, Hewett J, Farley J, Irimia D, Li Y, Wiche G, Sonnenberg A, Breakefield XO. 2008.
TorsinA binds the KASH domain of nesprins and participates in linkage between nuclear envelope and
cytoskeleton. Journal of Cell Science 121:3476–3486. doi: 10.1242/jcs.029454
Olivares AO, Baker TA, Sauer RT. 2016. Mechanistic insights into bacterial AAA+ proteases and protein-
remodelling machines. Nature Reviews Microbiology 14:33–44. doi: 10.1038/nrmicro.2015.4
Otwinowski Z, Minor W. 1997. [20] Processing of X-ray diffraction data collected in oscillation mode. Methods in
Enzymology. Elsevier. p 307–326. doi: 10.1016/S0076-6879(97)76066-X
Ozelius LJ, Hewett JW, Page CE, Bressman SB, Kramer PL, Shalish C, de Leon D, Brin MF, Raymond D, Corey
DP, Fahn S, Risch NJ, Buckler AJ, Gusella JF, Breakefield XO. 1997. The early-onset torsion dystonia gene
(DYT1) encodes an ATP-binding protein. Nature Genetics 17:40–48. doi: 10.1038/ng0997-40
Rose AE, Brown RS, Schlieker C. 2015. Torsins: not your typical AAA+ ATPases. Critical Reviews in Biochemistry
and Molecular Biology 50:532–549. doi: 10.3109/10409238.2015.1091804
Sauer RT, Baker TA. 2011. AAA+ proteases: ATP-fueled machines of protein destruction. Annual Review of
Biochemistry 80:587–612. doi: 10.1146/annurev-biochem-060408-172623
Sosa BA, Demircioglu FE, Chen JZ, Ingram J, Ploegh HL, Schwartz TU. 2014. How lamina-associated polypeptide
1 (LAP1) activates Torsin. eLife 3:e03239. doi: 10.7554/eLife.03239
So¨ding J, Biegert A, Lupas AN. 2005. The HHpred interactive server for protein homology detection and
structure prediction. Nucleic Acids Research 33:W244–W248. doi: 10.1093/nar/gki408
Turner EM, Brown RS, Laudermilch E, Tsai PL, Schlieker C. 2015. The Torsin Activator LULL1 Is Required for
Efficient Growth of Herpes Simplex Virus 1. Journal of Virology 89:8444–8452. doi: 10.1128/JVI.01143-15
Tusna´dy GE, Simon I. 2001. The HMMTOP transmembrane topology prediction server. Bioinformatics 17:849–
850. doi: 10.1093/bioinformatics/17.9.849
Vander Heyden AB, Naismith TV, Snapp EL, Hodzic D, Hanson PI, Hanson PI. 2009. LULL1 retargets TorsinA to
the nuclear envelope revealing an activity that is impaired by the DYT1 dystonia mutation. Molecular Biology of
the Cell 20:2661–2672. doi: 10.1091/mbc.E09-01-0094
Vulinovic F, Lohmann K, Rakovic A, Capetian P, Alvarez-Fischer D, Schmidt A, Weißbach A, Erogullari A, Kaiser
FJ, Wiegers K, Ferbert A, Rolfs A, Klein C, Seibler P. 2014. Unraveling cellular phenotypes of novel TorsinA/
TOR1A mutations. Human Mutation 35:1114–1122. doi: 10.1002/humu.22604
Waterhouse AM, Procter JB, Martin DM, Clamp M, Barton GJ. 2009. Jalview Version 2–a multiple sequence
alignment editor and analysis workbench. Bioinformatics 25:1189–1191. doi: 10.1093/bioinformatics/btp033
Wendler P, Ciniawsky S, Kock M, Kube S. 2012. Structure and function of the AAA+ nucleotide binding pocket.
Biochimica Et Biophysica Acta 1823:2–14. doi: 10.1016/j.bbamcr.2011.06.014
White SR, Lauring B. 2007. AAA+ ATPases: achieving diversity of function with conserved machinery. Traffic 8:
1657–1667. doi: 10.1111/j.1600-0854.2007.00642.x
Zeymer C, Barends TR, Werbeck ND, Schlichting I, Reinstein J. 2014. Elements in nucleotide sensing and
hydrolysis of the AAA+ disaggregation machine ClpB: a structure-based mechanistic dissection of a molecular
motor. Acta Crystallographica Section D Biological Crystallography 70:582–595. doi: 10.1107/
S1399004713030629
Zhao C, Brown RSH, Chase AR, Eisele MR, Schlieker C. 2013. Regulation of Torsin ATPases by LAP1 and LULL1.
PNAS 110:E1545–E1554. doi: 10.1073/pnas.1300676110
Zhu L, Millen L, Mendoza JL, Thomas PJ. 2010. A unique redox-sensing sensor II motif in TorsinA plays a critical
role in nucleotide and partner binding. Journal of Biological Chemistry 285:37271–37280. doi: 10.1074/jbc.
M110.123471
Zhu L, Wrabl JO, Hayashi AP, Rose LS, Thomas PJ. 2008. The torsin-family AAA+ protein OOC-5 contains a
critical disulfide adjacent to Sensor-II that couples redox state to nucleotide binding. Molecular Biology of the
Cell 19:3599–3612. doi: 10.1091/mbc.E08-01-0015
Zirn B, Grundmann K, Huppke P, Puthenparampil J, Wolburg H, Riess O, Muller U. 2008. Novel TOR1A mutation
p.Arg288Gln in early-onset dystonia (DYT1). Journal of Neurology, Neurosurgery & Psychiatry 79:1327–1330.
doi: 10.1136/jnnp.2008.148270
Demircioglu et al. eLife 2016;5:e17983. DOI: 10.7554/eLife.17983 14 of 14
Research article Biophysics and Structural Biology Cell Biology