Article

Insight into the Structure of the ‘‘Unstructured’’ Tau

Protein
Graphical Abstract Authors
Konstantin I. Popov,
Karl A.T. Makepeace,
Evgeniy V. Petrotchenko,
Nikolay V. Dokholyan,
Christoph H. Borchers

Correspondence
dokh@psu.edu (N.V.D.),
christoph.borchers@mcgill.ca (C.H.B.)

In Brief
Short-distance crosslinking constraint-
guided all-atom discrete molecular
dynamics simulations (CL-DMD) was
used to predict the conformational
ensemble of the full-length tau protein in
solution. Tau in solution was found to be a
fairly compact globular protein that
contains persistent structural features in
the regions important for pathological
aggregation.

Highlights
d 441-residue tau protein structure was solved by all-atom
DMD simulations

d Structural proteomics experimental data were used in protein

structure prediction

d Tau is a rather compact conformational ensemble with

distinct subdomain topology

d Extensive b sheet structure was detected in the aggregation-

prone R1-R4 repeat regions

Popov et al., 2019, Structure 27, 1–6

November 5, 2019 ª 2019 Published by Elsevier Ltd.
https://doi.org/10.1016/j.str.2019.09.003
 Please cite this article in press as: Popov et al., Insight into the Structure of the ‘‘Unstructured’’ Tau Protein, Structure (2019), https://doi.org/10.1016/
j.str.2019.09.003

Structure

Article

Insight into the Structure

of the ‘‘Unstructured’’ Tau Protein
Konstantin I. Popov,1,7 Karl A.T. Makepeace,2,7 Evgeniy V. Petrotchenko,3 Nikolay V. Dokholyan,4,*
and Christoph H. Borchers2,3,5,6,8,*
1Department of Biochemistry and Biophysics, University of North Carolina, Chapel Hill, NC 27599, USA
2University of Victoria-Genome British Columbia Proteomics Centre, 3101-4464 Markham Street, Vancouver Island Technology Park,
Victoria, BC V8Z7X8, Canada
3Segal Cancer Proteomics Centre, Lady Davis Institute, Jewish General Hospital, McGill University, Montreal, QC, H3T 1E2, Canada
4Department of Pharmacology, Department of Biochemistry & Molecular Biology, Penn State College of Medicine, Hershey, PA 17033, USA
5Department of Biochemistry and Microbiology, University of Victoria, Petch Building, Room 270d, 3800 Finnerty Road, Victoria, BC V8P 5C2,

Canada
6Gerald Bronfman Department of Oncology, Jewish General Hospital, McGill University, Montreal, QC, H3T 1E2, Canada
7These authors contributed equally
8Lead Contact

*Correspondence: dokh@psu.edu (N.V.D.), christoph.borchers@mcgill.ca (C.H.B.)

https://doi.org/10.1016/j.str.2019.09.003

SUMMARY to be involved in the pathogenesis of neurodegenerative dis-

Combining structural proteomics experimental data
with computational methods is a powerful tool for
protein structure prediction. Here, we apply a recently
developed approach for de novo protein structure
determination based on the incorporation of short-
distance crosslinking data as constraints in discrete
molecular dynamics simulations (CL-DMD), for the
determination of the conformational ensemble of tau
protein in solution. The predicted structures were
in agreement with surface modification and long-
distance crosslinking data. Tau in solution was found
as an ensemble of rather compact globular conforma-
tions with distinct topology, inter-residue contacts,
and a number of transient secondary-structure ele-
ments. Regions important for pathological aggrega-
tion consistently were found to contain b strands.
The determined structures are compatible with the
tau protein in solution being a molten globule at
near-ground state with persistent residual structural
features which we were able to capture by CL-DMD.
The predicted structure may facilitate an understand-
ing of the misfolding and oligomerization pathways of
the tau protein.
INTRODUCTION
The tau protein is involved in the pathogenesis of misfolding-
related neurodegenerative diseases, including Alzheimer dis-
ease (Beharry et al., 2014; Frost et al., 2015). A misfolding event
leads to the formation of oligomers and eventually amyloid
tangles. The former are believed to be toxic and eventually
lead to the death of neuronal cells (Walsh et al., 2002). Moreover,
a prion-like spread of this aggregation via the conversion of
native protein molecules by toxic oligomers has been suggested
eases (Clavaguera et al., 2015). Native tau is considered to be
an intrinsically disordered protein, although evidence of some
globular structure can be found in the literature (Ramachandran
and Udgaonkar, 2013). Thus, the native structure of tau in solu-
tion may serve as a basis for understanding the misfolding and
oligomerization pathways.
Traditional all-atom molecular dynamics simulations of the tau
structure are hampered by the large size of the full-length tau pro-
tein (441 residues) (Nath et al., 2012). A number of biophysical
methods, such as NMR, electron paramagnetic resonance, fluo-
rescence resonance energy transfer (FRET), and small-angle
X-ray scattering—in combination with computational methods—
has been applied to the study of intrinsically disordered proteins,
including the structure of tau in solution (Jeganathan et al., 2006;
Mylonas et al., 2008; Nath et al., 2012; Zabik et al., 2017). A
tau conformational ensemble has been proposed based on
long-distance paramagnetic relaxation enhancement NMR or
single-molecule FRET data and molecular modeling (Mukrasch
et al., 2009). Nuclear Overhauser effect-derived short-distance
restrained molecular dynamics simulation are traditionally used
for the determination and refinement of protein structures by
NMR (Kaptein et al., 1985; Torda et al., 1990). Incorporation of
the crosslinking data as experimental distance constraints into
molecular dynamics protein structure modeling is an alternative
promising approach for protein structure determination (Peter


and Cerný, 2019). Recently, we developed a method for deter-
mining protein structures, called short-distance crosslinking
constraint-guided discrete molecular dynamics simulations (CL-
DMD), in which the folding process is guided by short-distance
experimental constraints that are incorporated into the DMD
force-field energy function (Brodie et al., 2017). Adding short-dis-
tance constraints to DMD simulations results in a reduction of the
possible conformational space and allows simulations to
converge more rapidly to folded protein conformations (Chen
et al., 2007). We have tested this approach on well-structured pro-
teins and, recently, for the prediction of the conformational
ensemble of the 140-residue-long intrinsically disordered protein,
a-synuclein (Brodie et al., 2019). The conformational flexibility of

Structure 27, 1–6, November 5, 2019 ª 2019 Published by Elsevier Ltd. 1

 Please cite this article in press as: Popov et al., Insight into the Structure of the ‘‘Unstructured’’ Tau Protein, Structure (2019), https://doi.org/10.1016/
j.str.2019.09.003
A B
intrinsically disordered proteins, such as a-synuclein and tau,
brings additional challenges to the computational process
(Bonomi et al., 2017), because, in these cases, proteins exist as
a collection of inter-converting conformational states, and the
crosslinking data represent multiple conformations of a protein
rather than a single structure.
The Medusa force field (Ding et al., 2006, 2008; Shirvany-
ants et al., 2012), which is utilized in DMD simulations, is dis-
cretized to mimic continuous inter-atom potentials, and can
readily integrate any additional potentials, such as pairwise
distance constraints (Chen et al., 2007; Proctor et al., 2011)
and solvent accessibility information (Ding et al., 2012; Dixon
et al., 2004). Using CL-DMD simulations, we generate confor-
mational ensembles that satisfy the optimal number of con-
straints, thereby naturally helping to resolve possibly conflict-
ing experimentally derived constraints. Recently, we showed
that CL-DMD simulations are a viable computational platform
for the structural analysis of intrinsically disordered proteins,
and a-synuclein in particular (Brodie et al., 2019; Szöllo } si
et al., 2014).
Here, we extend the CL-DMD approach to determine the
conformational ensembles of the full-length tau protein in
solution. During this process, tau was crosslinked with a panel
of short-range crosslinkers (spacer length <7 Å), the crosslinked
proteins were enzymatically digested, the crosslinked residues
were determined by liquid chromatography-tandem mass spec-
trometry (LC-MS/MS) analysis, and the resulting information on
inter-residue distances was introduced into the DMD force field
as external constraints. To experimentally validate the predicted
structures, we analyzed tau using surface modification (SM) and
long-distance crosslinking (LD-CL).
RESULTS AND DISCUSSION
To determine the tau structure, we used our recently developed
approach of crosslinking constraint-guided discrete molecular
dynamics, CL-DMD (Brodie et al., 2017). The CL-DMD proced-
ure uses experimental inter-residue short-distance crosslinking
constraints for guiding DMD simulations, thus allowing to
achieve modeling of the protein folding in reasonable time.
In brief, tau was crosslinked with a panel of short-range re-
agents succinimidyl 4,40 -azipentanoate (SDA) (Brodie et al.,
2015), DSG, and DSA. SDA is a hetero-bifunctional amino
group-reactive and photoreactive reagent, spacer length 5 Å,
2 Structure 27, 1–6, November 5, 2019
Figure 1. Conformational Ensemble of Native
Tau in Solution as Determined by CL-DMD
(A) Conformational ensemble of tau. Representatives
of the major clusters are aligned and colored gray.
Lowest-energy centroid is shown in cartoon repre-
sentation and colored N-to-C from blue-to-red.
(B) Subdomain topology of the tau conformational
ensemble. Residues are colored as in (A).
and DSG and DSA are homobifunctional
amino group-reactive crosslinkers with
spacer lengths of 6 and 7 Å, respectively.
Crosslinked proteins were separated by
SDS-PAGE, the monomer band was in-gel
digested with proteinase K or trypsin proteolytic enzymes, and
the digest was analyzed by LC-MS/MS to identify crosslinked
peptides (Table S1). We used the monomer band of the cross-
linked tau to exclude potential inter-protein crosslinks from the
analysis. The distances between crosslinked residues are based
on the length of the crosslinker reagents (Table S1), and were
introduced as constraints into the DMD potentials (see the
STAR Methods and Brodie et al., 2017, for additional details).
In addition, tau was characterized by SM and LD-CL. SM exper-
imental data were extracted from crosslinking experiments using
non-selective photoreactive diazirine-based reagent SDA to
determine the characteristics of the residues as exposed or
buried (Table S2). LD-CL was used to estimate the overall protein
topology (Table S3).
To obtain information on the global folding of tau, we per-
formed clustering analysis on the lowest-energy structures
obtained during CL-DMD simulations (see the STAR
Methods). In ordered proteins, the lowest-energy structures
are usually represented by only one or few states, whose con-
formations are close to the corresponding native structures.
In contrast, disordered proteins are usually represented by a
broader variety of distinct structures, reflecting the conforma-
tional freedom of the intrinsically disordered protein (Pauwels
et al., 2017). In our case, the lowest-energy structures forming
the conformational ensemble for tau are shown as overlays in
Figure 1A. They can be described as rather compact globular
structures with a common general topology, containing a
number of secondary-structure elements. A distinct topology
which is shared between all of these conformers can be
observed. Four subdomains can be recognized in the struc-
tures, which arranged in a tetrahedral fashion (Figure 1B).
We did not find the C-terminal portion of the molecule to be
in close proximity to the N-terminal portion with extensive
contacts, as was suggested earlier (Mukrasch et al., 2009).
In contrast, we found it to be confined on both sides by two
intermediate subdomains.
In any structural proteomics experiment, some of the experi-
mentally derived constraints may conflict with each other
because proteins exist in multiple conformations. During DMD
simulations, the system is driven to satisfy the maximum
possible number of constraints, so the generated ensemble of
structures satisfies subsets of consistent experimental con-
straints. The representative models are then selected by per-
forming geometrical clustering of this ensemble and selecting
 Please cite this article in press as: Popov et al., Insight into the Structure of the ‘‘Unstructured’’ Tau Protein, Structure (2019), https://doi.org/10.1016/
j.str.2019.09.003
Figure 2. Original and Relaxed Structures of the Lowest-Energy
Centroid
The starting conformer is shown in rainbow colors and the relaxed without
crosslinking distance constraints conformer is shown in gray.
the centroids of the most populated clusters. Depending on the
conformation and the number of constraints that are satisfied,
these structures will have various energy scores assigned by
the force field (Ding et al., 2008; Yin et al., 2008). This allows
us to take into account both the nature of the conformational
changes of a protein in solution (i.e., while the structure of the
protein is ‘‘breathing’’ and transitioning between conformations),
as well as mitigate possible errors that may have occurred during
the determination of the experimental constraints (Bonomi et al.,
2017). Nevertheless, we found that the lowest-energy con-
formers satisfy most of the experimental distance constraints,
while the models with higher energy satisfy a lower number of
constraints. A comparison of predicted structures within an
ensemble revealed that the conformational changes to the pro-
tein structure were mainly due to internal rearrangements within
the N-, C-terminal, and intermediate subdomains. One could
argue that the short-distance constraints used in the simulations
might drive the structures to more compact states. To address
this concern, we have removed all of the experimental con-
straints from the obtained structures, and have then run uncon-
strained all-atom simulations using GROMACS 2018 and
CHARMM36 force field for an additional 54 ns in 3 replicates
(see the STAR Methods). During first 4 ns each system was
A B
1 50 100 150 200 250 300 350 400
0 441
subjected to the simulated annealing procedure, where both
the protein and the solvent were four times cyclically heated
up from initial 300 K temperature to 325 K for 0.5 ns and then
cooled back to 300 K for another 0.5 ns. After that, regular
molecular dynamics simulations were performed for 50 ns. The
replicate trajectories were subjected to clustering, and the cen-
troids of the new most populated clusters were selected as our
new relaxed models. We found that the relaxed structures had
generally the same conformation (3 Å root-mean-square devia-
tion between the lowest constrained and unconstrained energy
states), with only a slight relaxation of the more flexible regions,
which suggests that the experimental constraints did not signif-
icantly bias the models toward artificially overcompacted states.
For simplicity, only the relaxed structure of the first centroid is
presented in Figure 2.
Some secondary-structure elements were observed in the con-
formers of the ensemble (Figure 3). The extent of the predicted
secondary-structure motifs in the lowest-energy conformer was
higher than previously proposed (Mukrasch et al., 2009), but, as
we showed in the case of a-synuclein (Brodie et al., 2019), this
probably reflects the presence of transient secondary-structure
elements. Prediction of the secondary-structure content from
CD data (Mirbaha et al., 2018) using the BeStSel server (Micsonai
et al., 2018) (Figure S1) was in general agreement with the content
found in the lowest-energy centroid (Figure 3): 11% versus 7% of
helix and 21% versus 31% of antiparallel b structure for the model
and predictions, respectively.
The SM results were in agreement with the final tau structures
(Figure 4A). We used photoreactive SDA dead-end crosslinks
(the photoreactive part of the reagent is attached to the protein
residues and the NHS moiety of the reagent is hydrolyzed) as
quasi SM reagents. SDA is a non-selective diazirine-based
crosslinking reagent and has the potential to modify any protein
residue, although some reactivity preferences for diazirine-
based reagents have been reported (Ziemianowicz et al.,
2017). We have detected 155 modified protein residues (Table
S2). Most modifications were consistent with the lowest-energy
conformer structure (only 8 of the 155 modified residues in the
structure cannot be directly accessed from protein surface by
the modifying reagent) (Figure 4A). LD-CL, as we have shown
earlier, cannot be used directly in CL-DMD simulations but can
be employed for confirmation of the overall topology of the pro-
tein models (Brodie et al., 2017). We used DSS crosslinker
Figure 3. Short-Distance Crosslinks Used for
CL-DMD of Tau Protein in Solution
Lowest-energy conformer from the ensemble as in
Figure 1 is shown. (A) Short-distance crosslinks used
in the CL-DMD simulations are shown as lines con-
necting Ca atoms and as circular arcs on the diagram.
Red, green, and blue lines correspond to DSA, DSG,
and SDA crosslinks, respectively. (B) Amiloidogenic
R2, R3, and R4 peptides 275VQIINK280, 306VQIVYK311,
and 337VEVK340 are highlighted in orange. The V306-
F378 region found by cryo-EM forming cross-b
structure in fibrils (Fitzpatrick et al., 2017) is shown in
light orange.
Structure 27, 1–6, November 5, 2019 3
 Please cite this article in press as: Popov et al., Insight into the Structure of the ‘‘Unstructured’’ Tau Protein, Structure (2019), https://doi.org/10.1016/
j.str.2019.09.003
A B
(spacer length 11 Å) for LD-CL (Table S3). Fifty-two long-dis-
tance intra-protein DSS crosslinks were also in agreement with
the final lowest-energy conformer of the protein (Figure 4B). Pre-
viously reported long-distance DSS crosslinks (Mirbaha et al.,
2018) were also found to be in agreement with the obtained
structure (Figure 4B), with only one out of ten observed cross-
links exceeding 30 Å length (Table S3).
Overall, the tau conformational ensemble determined here
can be considered as a representation of a wide energy-funnel
characteristic of the molten globular state of disordered pro-
teins. For each particular representative within the ensemble,
CL-DMD was able to capture the structural features present.
This structure is in agreement with our earlier finding of resid-
ual native-like structural features in unfolded proteins (Ding
et al., 2005). This distribution of near-ground-state inter-con-
verting conformations may be the basis for the structural plas-
ticity of intrinsically disordered proteins (Pauwels et al., 2017),
and the existing persistent structural features can determine
the subsequent misfolding and oligomerization pathways.
The 2N4R isoform of tau protein used in this study contains
an N-terminal projection domain with N1 and N2 repeats, a
proline-rich domain, a microtubule binding region containing
four R1, R2, R3, and R4 repeats, and a C-terminal domain
(Ramachandran and Udgaonkar, 2013). The R1-R4 repeats
are reported to be involved in microtubule binding and are
important for aggregation, with the R3 and R4 repeats forming
a cross-b sheet core of paired helical filaments in vivo (Fitzpa-
trick et al., 2017). Interestingly, in our structure this region was
found to contain multiple b structure elements (Figure 3), which
is consistent with its tendency to form b structures in the ag-
gregates. In fact, seven out of eight b strands of the fibrils
core (the exception being the b2 strand) were already found
in to be in a b structural form in the lowest-energy conformer
(Figure 3). The central 347KDR349 turn was also found as a
loop connecting two b strands. Based on these observed
structural features, we hypothesize that the mechanism of
conformational change in this region (which leads to aggrega-
tion) is a ‘‘pulling away’’ of the amino acid 341–357 hairpin
from the core of the molecule, with further re-orientation of
the hydrogen bonding from intra-protein to inter-protein
cross-b bonding. Most of this region is buried in our structure,
which is also consistent with the idea that a dramatic rear-
rangement of the protein molecule needs to occur to initiate
pathological aggregation.
4 Structure 27, 1–6, November 5, 2019
Figure 4. Experimental Validation of the Tau
Structure with SM and LD-CL
(A) Side chains of residues that are found to be
modified as dead-ends by photoreactive moiety of
hydrolyzed SDA crosslinking reagent are shown as
spheres.
(B) LD-CL DSS crosslinks are shown as black lines.
Hyper-phosphorylation has been pro-
posed as a factor facilitating conversion
from the native to the pathological aggre-
gated form of tau. Tau can be phosphory-
lated at 30 sites by multiple kinases (Mair
et al., 2016). Hypothetically, phosphorylation
in this region—at least at residues S293, S324, and S356—may
enable the ‘‘loosening’’ and opening up of the structure, thereby
exposing the residues to the solvent and to a potentially interact-
ing tau molecule during aggregate formation.
In summary, a structural conformational ensemble of native
tau in solution was determined by CL-DMD using short-distance
constraints derived from experimentally observed inter-residue
crosslinks. The predicted structure was corroborated by SM
and LD-CL experimental data. The general folding pattern and
the transient secondary-structure motifs found in our predicted
structure of the tau protein may help to explain the structural
transitions of the molecule that occur during the pathological
conversion and oligomerization.
Conclusions
We have determined de novo the native structure of tau in solu-
tion by short-distance CL-DMD simulations, and validated this
structure with experimental data from SM and LD-CL experi-
ments. The predicted structure is represented by an ensemble
of rather compact globular conformations with transient second-
ary-structure elements. The region containing R1-R4 repeats
was found to contain multiple b structure elements, which is
consistent with the tendency of this region to form a cross-b
structure in the aggregates. The obtained structure can serve
as a starting point for analyzing the misfolding and oligomeriza-
tion of the tau protein.
STAR+METHODS
Detailed methods are provided in the online version of this paper
and include the following:
d KEY RESOURCES TABLE
d LEAD CONTACT AND MATERIALS AVAILABILITY
B Materials
d METHOD DETAILS
B Expression and Purification of the Tau Protein
B Crosslinking
B LC-MS/MS Analysis
B Surface Modification
B Discrete Molecular Dynamics Modeling
B Unconstrained Molecular Dynamics Modeling
d QUANTIFICATION AND STATISTICAL ANALYSIS
d DATA AND CODE AVAILABILITY
 Please cite this article in press as: Popov et al., Insight into the Structure of the ‘‘Unstructured’’ Tau Protein, Structure (2019), https://doi.org/10.1016/
j.str.2019.09.003
SUPPLEMENTAL INFORMATION
Supplemental Information can be found online at https://doi.org/10.1016/j.str.
2019.09.003.
ACKNOWLEDGMENTS
The University of Victoria-Genome British Columbia Proteomics Center is
grateful to Genome Canada and Genome British Columbia for financial
support through the Genomics Innovation Network (codes 204PRO for
operations and 214PRO for technology development) and the Genomics
Technology Platform (264PRO). C.H.B. would also like to thank the Natural
Sciences and Engineering Research Council of Canada of Canada (NSERC)
and the Leading Edge Endowment Fund for support. C.H.B. is also grateful
for support from the Segal McGill Chair in Molecular Oncology at McGill Uni-
versity (Montreal, QC, Canada), and for support from the Warren Y. Soper
Charitable Trust and the Alvin Segal Family Foundation to the Jewish General
Hospital (Montreal, QC, Canada). This work was also supported by NIH grant
R01GM080742 to N.V.D. N.V.D. also acknowledges support from NIH grants
R01GM114015 and R01GM123247.
AUTHOR CONTRIBUTIONS
Conceptualization, E.V.P., C.H.B., and N.V.D.; Methodology, E.V.P., C.H.B.,
N.V.D., K.I.P., and K.A.T.M.; Investigation, K.A.T.M. and K.I.P.; Writing – Orig-
inal Draft, E.V.P., K.A.T.M., and K.I.P.; Writing – Review & Editing, all authors;
Funding Acquisition and Supervision, C.H.B. and N.V.D.
DECLARATION OF INTERESTS
C.H.B. and E.V.P. are co-founders of Creative Molecules, Inc. The other
authors declare no competing interests.
Received: March 4, 2019
Revised: July 2, 2019
Accepted: September 12, 2019
Published: October 15, 2019
REFERENCES
Beharry, C., Cohen, L.S., Di, J., Ibrahim, K., Briffa-Mirabella, S., and Alonso,
A.e.C. (2014). Tau-induced neurodegeneration: mechanisms and targets.
Neurosci. Bull. 30, 346–358.
Best, R.B., Zhu, X., Shim, J., Lopes, P.E., Mittal, J., Feig, M., and Mackerell,
A.D. (2012). Optimization of the additive CHARMM all-atom protein force field
targeting improved sampling of the backbone f, c and side-chain c(1) and c(2)
dihedral angles. J. Chem. Theor. Comput. 8, 3257–3273.
Bonomi, M., Heller, G.T., Camilloni, C., and Vendruscolo, M. (2017). Principles
of protein structural ensemble determination. Curr. Opin. Struct. Biol. 42,
106–116.
Brodie, N.I., Makepeace, K.A., Petrotchenko, E.V., and Borchers, C. (2015).
Isotopically-coded short-range hetero-bifunctional photo-reactive cross-
linkers for studying protein structure. J. Proteomics 118, 12–20.
Brodie, N.I., Popov, K.I., Petrotchenko, E.V., Dokholyan, N.V., and Borchers,
C.H. (2017). Solving protein structures using short-distance cross-linking
constraints as a guide for discrete molecular dynamics simulations. Sci.
Adv. 3, e1700479.
Brodie, N.I., Popov, K.I., Petrotchenko, E.V., Dokholyan, N.V., and Borchers,
C.H. (2019). Conformational ensemble of native a-synuclein in solution as
determined by short-distance crosslinking constraint-guided discrete molec-
ular dynamics simulations. PLoS Comput. Biol. 15, e1006859.
Chen, Y., Ding, F., and Dokholyan, N.V. (2007). Fidelity of the protein structure
reconstruction from inter-residue proximity constraints. J. Phys. Chem. B 111,
7432–7438.
Clavaguera, F., Hench, J., Goedert, M., and Tolnay, M. (2015). Invited review:
prion-like transmission and spreading of tau pathology. Neuropathol. Appl.
Neurobiol. 41, 47–58.
Ding, F., Dokholyan, N.V., and Shakhnovich, E. (2006). Emergence of protein
fold families through rational design. PLoS Comput. Biol. 2, e85.
Ding, F., Furukawa, Y., Nukina, N., and Dokholyan, N.V. (2012). Local unfolding
of Cu, Zn superoxide dismutase monomer determines the morphology of
fibrillar aggregates. J. Mol. Biol. 421, 548–560.
Ding, F., Jha, R.K., and Dokholyan, N.V. (2005). Scaling behavior and structure
of denatured proteins. Structure 13, 1047–1054.
Ding, F., Tsao, D., Nie, H., and Dokholyan, N.V. (2008). Ab initio folding of pro-
teins with all-atom discrete molecular dynamics. Structure 16, 1010–1018.
Dixon, R.D., Chen, Y., Ding, F., Khare, S.D., Prutzman, K.C., Schaller, M.D.,
Campbell, S.L., and Dokholyan, N.V. (2004). New insights into FAK signaling
and localization based on detection of a FAT domain folding intermediate.
Structure 12, 2161–2171.
Fitzpatrick, A.W.P., Falcon, B., He, S., Murzin, A.G., Murshudov, G., Garringer,
H.J., Crowther, R.A., Ghetti, B., Goedert, M., and Scheres, S.H.W. (2017).
Cryo-EM structures of tau filaments from Alzheimer’s disease. Nature 547,
185–190.
Frost, B., Götz, J., and Feany, M.B. (2015). Connecting the dots between tau
dysfunction and neurodegeneration. Trends Cell Biol. 25, 46–53.
Hess, B., Kutzner, C., van der Spoel, D., and Lindahl, E. (2008). GROMACS 4:
algorithms for highly efficient, load-balanced, and scalable molecular simula-
tion. J. Chem. Theor. Comput. 4, 435–447.
Hoopmann, M.R., Zelter, A., Johnson, R.S., Riffle, M., MacCoss, M.J., Davis,
T.N., and Moritz, R.L. (2015). Kojak: efficient analysis of chemically cross-
linked protein complexes. J. Proteome Res. 14, 2190–2198.
Jeganathan, S., von Bergen, M., Brutlach, H., Steinhoff, H.J., and Mandelkow,
E. (2006). Global hairpin folding of tau in solution. Biochemistry 45, 2283–2293.
€ll, L., Canterbury, J.D., Weston, J., Noble, W.S., and MacCoss, M.J. (2007).
Ka
Semi-supervised learning for peptide identification from shotgun proteomics
datasets. Nat. Methods 4, 923.
Kaptein, R., Zuiderweg, E.R., Scheek, R.M., Boelens, R., and van Gunsteren,
W.F. (1985). A protein structure from nuclear magnetic resonance data. lac
repressor headpiece. J. Mol. Biol. 182, 179–182.
Mair, W., Muntel, J., Tepper, K., Tang, S., Biernat, J., Seeley, W.W., Kosik,
K.S., Mandelkow, E., Steen, H., and Steen, J.A. (2016). FLEXITau: quantifying
post-translational modifications of tau protein in vitro and in human disease.
Anal. Chem. 88, 3704–3714.
Micsonai, A., Wien, F., Bulyáki, É., Kun, J., Moussong, É., Lee, Y.H., Goto, Y.,
Réfrégiers, M., and Kardos, J. (2018). BeStSel: a web server for accurate
protein secondary structure prediction and fold recognition from the circular
dichroism spectra. Nucleic Acids Res. 46, W315–W322.
Mirbaha, H., Chen, D., Morazova, O.A., Ruff, K.M., Sharma, A.M., Liu, X.,
Goodarzi, M., Pappu, R.V., Colby, D.W., Mirzaei, H., et al. (2018). Inert and
seed-competent tau monomers suggest structural origins of aggregation.
Elife 7, e36584.
Mukrasch, M.D., Bibow, S., Korukottu, J., Jeganathan, S., Biernat, J.,
Griesinger, C., Mandelkow, E., and Zweckstetter, M. (2009). Structural poly-
morphism of 441-residue tau at single residue resolution. PLoS Biol. 7, e34.
Mylonas, E., Hascher, A., Bernadó, P., Blackledge, M., Mandelkow, E., and
Svergun, D.I. (2008). Domain conformation of tau protein studied by solution
small-angle X-ray scattering. Biochemistry 47, 10345–10353.
Nath, A., Sammalkorpi, M., DeWitt, D.C., Trexler, A.J., Elbaum-Garfinkle, S.,
O’Hern, C.S., and Rhoades, E. (2012). The conformational ensembles of
alpha-synuclein and tau: combining single-molecule FRET and simulations.
Biophys. J. 103, 1940–1949.
Okamoto, Y. (2004). Generalized-ensemble algorithms: enhanced sampling
techniques for Monte Carlo and molecular dynamics simulations. J. Mol.
Graph. Model. 22, 425–439.
Pauwels, K., Lebrun, P., and Tompa, P. (2017). To be disordered or not to be
disordered: is that still a question for proteins in the cell? Cell. Mol. Life Sci. 74,
3185–3204.
Perez-Riverol, Y., Csordas, A., Bai, J., Bernal-Llinares, M., Hewapathirana, S.,
Kundu, D.J., Inuganti, A., Griss, J., Mayer, G., Eisenacher, M., et al. (2019). The
Structure 27, 1–6, November 5, 2019 5
 Please cite this article in press as: Popov et al., Insight into the Structure of the ‘‘Unstructured’’ Tau Protein, Structure (2019), https://doi.org/10.1016/
j.str.2019.09.003
PRIDE database and related tools and resources in 2019: improving support
for quantification data. Nucleic Acids Res. 47, D442–D450.


Peter, E.K., and Cerný, J. (2019). A hybrid Hamiltonian for the
accelerated sampling along experimental restraints. Int. J. Mol. Sci.
20, E370.
Proctor, E.A., Ding, F., and Dokholyan, N.V. (2011). Discrete molecular dy-
namics. Wiley Interdiscip. Rev. Comput. Mol. Sci. 1, 80–92.
Pronk, S., Pall, S., Schulz, R., Larsson, P., Bjelkmar, P., Apostolov, R., Shirts,
M.R., Smith, J.C., Kasson, P.M., van der Spoel, D., et al. (2013). GROMACS
4.5: a high-throughput and highly parallel open source molecular simulation
toolkit. Bioinformatics 29, 845–854.
Ramachandran, G., and Udgaonkar, J.B. (2013). Mechanistic studies unravel
the complexity inherent in tau aggregation leading to Alzheimer’s disease
and the tauopathies. Biochemistry 52, 4107–4126.
Shevchenko, A., Tomas, H., Havlis, J., Olsen, J.V., and Mann, M. (2006). In-gel
digestion for mass spectrometric characterization of proteins and proteomes.
Nat. Protoc. 1, 2856–2860.
Shirvanyants, D., Ding, F., Tsao, D., Ramachandran, S., and Dokholyan,
N.V. (2012). Discrete molecular dynamics: an efficient and versatile simu-
lation method for fine protein characterization. J. Phys. Chem. B 116,
8375–8382.
6 Structure 27, 1–6, November 5, 2019
} si, D., Horváth, T., Han, K.H., Dokholyan, N.V., Tompa, P., Kalmár, L.,
Szöllo
and Hegedu }s, T. (2014). Discrete molecular dynamics can predict helical pre-
structured motifs in disordered proteins. PLoS One 9, e95795.
Torda, A.E., Scheek, R.M., and van Gunsteren, W.F. (1990). Time-averaged
nuclear Overhauser effect distance restraints applied to tendamistat. J. Mol.
Biol. 214, 223–235.
Walsh, D.M., Klyubin, I., Fadeeva, J.V., Cullen, W.K., Anwyl, R., Wolfe, M.S.,
Rowan, M.J., and Selkoe, D.J. (2002). Naturally secreted oligomers of amyloid
beta protein potently inhibit hippocampal long-term potentiation in vivo.
Nature 416, 535–539.
Yin, S., Biedermannova, L., Vondrasek, J., and Dokholyan, N.V. (2008).
MedusaScore: an accurate force field-based scoring function for virtual drug
screening. J. Chem. Inf. Model. 48, 1656–1662.
Yuzwa, S.A., Yadav, A.K., Skorobogatko, Y., Clark, T., Vosseller, K., and
Vocadlo, D.J. (2011 Mar). Mapping O-GlcNAc modification sites on tau and gen-
eration of a site-specific O-GlcNAc tau antibody. Amino Acids. 40 (3), 857–868.
Zabik, N.L., Imhof, M.M., and Martic-Milne, S. (2017). Structural evaluations of
tau protein conformation: methodologies and approaches. Biochem. Cell Biol.
95, 338–349.
Ziemianowicz, D.S., Bomgarden, R., Etienne, C., and Schriemer, D.C. (2017).
Amino acid insertion frequencies arising from photoproducts generated using
aliphatic diazirines. J. Am. Soc. Mass Spectrom. 28, 2011–2021.
 Please cite this article in press as: Popov et al., Insight into the Structure of the ‘‘Unstructured’’ Tau Protein, Structure (2019), https://doi.org/10.1016/
j.str.2019.09.003

STAR+METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER

Chemicals, Peptides, and Recombinant Proteins
Acetonitrile (HPLC), Fisher Chemical Thermo Scientific Cat#: A9984
Ammonium bicarbonate Sigma-Aldrich SKU#: A6141
Bond-Breaker TCEP Solution, Neutral pH Thermo Scientific Cat#: 77720
CBDPS-H8/D8 CyanurBiotinDimercaptoPropionylSuccinimide Creative Molecules Inc. Cat#: 014S
cOmplete, EDTA-free Protease Inhibitor Roche SKU#: 4693132001
Cocktail Tablets provided in EASYpacks
Dimethyl sulphoxide Caledon Cat#: 4100-3
DL-Dithiothreitol Sigma-Aldrich Cat#: D0632
DSA-12C6/13C6 DiSuccinimidylAdipate Creative Molecules Inc. Cat#: 013SC
DSG-H6/D6 DiSuccinimidylGlutarate Creative Molecules Inc. Cat#: 010S
Formic Acid, 88% (Certified ACS) Fisher Scientific Cat#: A118P
Hydrochloric acid BDH Cat#: BDH3026
Imidazole Bio Basic Cat#: IB0277
Isopropyl b-D-thiogalactoside Sigma-Aldrich SKU#: I6758
Ni-NTA Agarose QIAGEN Cat#: 1018244
NuPAGE LDS Sample Buffer (4X) Invitrogen Cat#: NP0008
NuPAGE MOPS SDS Running Buffer (20X) Invitrogen Cat#: NP0001
Phenylmethylsulfonyl fluoride Roche SKU#: 10837091001
Phosphate buffered saline (PBS), pH 7.4 (10X stock) Invitrogen Cat#: AM9625
SDA (NHS-Diazirine) (succinimidyl 4,4’-azipentanoate) Thermo Scientific Cat#: 26167
Sequencing Grade Modified Trypsin Promega Cat#: V511C
Sodium hydroxide Bio Basic Cat#: SB6789
Trifluoroacetic Acid (TFA), Sequencing grade Thermo Scientific Cat#: 28901
Water, Optima LC/MS Grade, Fisher Chemical Thermo Scientific Cat#: W64
Deposited Data
Cross-linking Mass Spectrometry Data This paper Website: http://www.ebi.ac.uk/pride Project
Webpage: http://www.ebi.ac.uk/pride/archive/
projects/PXD015044
Project ProteomeXchange accession:
PXD015044
Tau structure This paper PDBDEV_00000033
Experimental Models: Organisms/Strains
BL21(DE3) Competent E. coli New England Biolabs Cat#: C2527I
Recombinant DNA
pET28a plasmid vector containing a synthetic gene for Dr. David Vocadlo; N/A
human 2N4R tau with additional 21-reisdue N-terminal Yuzwa et al., 2011
6x-His tag
Software and Algorithms
CL-DMD Molecules in Action Website: http://www.moleculesinaction.
com/pdmd.html
GROMACS Hess et al., 2008 https://doi.org/10.1021/ct700301q
Kojak (v. 1.5.5) Institute for Systems Biology; Website: http://www.kojak-ms.org/
Hoopmann et al., 2015
Microsoft Excel 2010 Microsoft N/A
Percolator (v. 2.08) €ll et al., 2007
Ka Website: http://percolator.ms/
Xcalibur (v. 2.2) Thermo Scientific N/A
(Continued on next page)

Structure 27, 1–6.e1–e4, November 5, 2019 e1

 Please cite this article in press as: Popov et al., Insight into the Structure of the ‘‘Unstructured’’ Tau Protein, Structure (2019), https://doi.org/10.1016/
j.str.2019.09.003

Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
Other
5415 D Centrifuge Eppendorf Model: 5415 D
Amicon Ultra-15 Centrifugal Filter Units (10,000 NMWL) Millipore Cat#: UFC901008
Biorad Model 200/2.0 Electrophoresis Power Supply Bio-Rad Laboratories N/A
Centrifuge 5810 R (15 amp) Eppendorf Cat#: 022625501
EASY-nLC 1000 Liquid Chromatography System ThermoFisher Scientific Cat#: LC120
EASY-nLC II Liquid Chromatography System ThermoFisher Scientific Cat#: LC110
Forma 900 Series -86 C Upright Ultra-Low Temperature Thermo Scientific N/A
Freezers
FPLC Liquid Chromatography Controller LCC-500 Plus Pharmacia N/A
FPLC LKB FRAC-100 Fraction Collector Pharmacia N/A
FPLC Mixer module, 24V, 50-60 Hz, 5 Mpa Pharmacia N/A
FPLC Motor Valve MV-7, 24V DC, 10 Mpa Pharmacia N/A
FPLC Pump P-500 Pharmacia N/A
FPLC Single Path Monitor UV-1 Control Unit Pharmacia N/A
FPLC Single Path Monitor UV-1 Optical Unit Pharmacia N/A
Gastight Syringe, 1 mL Hamilton Model#: 1001
Handheld UV Lamp Model UVGL-58 UVP Inc. Cat#: 95-0007-05
IntegraFrit New Objective N/A
JA-20 Fixed-Angle Rotor Beckman N/A
L8-M Ultracentrifuge Beckman N/A
Magic C18AQ 100 Å, 5-mm pore size Michrom Cat#: H255
MaxQ 6000 Incubated/Refrigerated Stackable Shaker Thermo Scientific Cat#: SHKE6000
Mini Gel Tank Invitrogen Cat#: A25977
Novaspec III+ Spectrophotometer Biochrom Cat#: 80-2120-40
NuPAGE 4-12% Bis-Tris Protein Gels, 1.0 mm, 12-well Invitrogen Cat#: NP0322BOX
Orbitrap Fusion Mass Spectrometer ThermoFisher Scientific N/A
Orbitrap Velos Pro Mass Spectrometer ThermoFisher Scientific N/A
S-3000 Sonicator Misonix Inc N/A
Superdex 200 10/300 GL GE Healthcare Life Sciences Cat#: 17517501
ThermoMixer C Eppendorf N/A

LEAD CONTACT AND MATERIALS AVAILABILITY

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Christoph
H. Borchers (christoph.borchers@mcgill.ca).

Materials
All reagents were from Sigma-Aldrich except the following: CBDPS, DSA, DSG crosslinking reagents (Creative Molecules Inc.), SDA
(ThermoFisher Scientific).

METHOD DETAILS

Expression and Purification of the Tau Protein

The pET28a plasmid vector was received as a gift from Prof. Dr. David Vocadlo, and containing a synthetic gene corresponding to
the human 2N4R tau isoform (441-residue isoform) with an additional 21-residue N-terminal fusion tag containing 6x-His and a
thrombin cleavage site (MGSSHHHHHHSSGLVPRGSHM). The plasmid was transfected into E. coli BL21(DE3) bacteria for
protein expression. Cell cultures were grown in lysogeny broth (LB) media at 37 C, with 150 rpm shaking, to an optical density
of 0.8-1.0 at 600 nm. Protein expression was induced by the addition of IPTG to give a final concentration of 0.5 or 1.0 mM. The
tau protein was then overexpressed overnight (16 hours) or for 4 hours at ambient room temperature (16-20 C), with 150 rpm
shaking.

e2 Structure 27, 1–6.e1–e4, November 5, 2019

 Please cite this article in press as: Popov et al., Insight into the Structure of the ‘‘Unstructured’’ Tau Protein, Structure (2019), https://doi.org/10.1016/
j.str.2019.09.003

Cells were harvested by centrifugation at 4000 rcf for 10 min at 4 C. The supernatant was discarded and the cells were resus-
pended in 30 mL of ice-cold phosphate buffered saline (PBS) at pH 7.4. The centrifugation step was repeated, the supernatant
was discarded, and the pelleted cells were then frozen at -80 C until needed. Frozen cell pellets were thawed and resuspended
in 30 mL of lysis buffer (20 mM sodium phosphate, 500 mM sodium chloride, 5 mM imidazole, pH 7.4), and two Roche complete
protease inhibitor cocktail tablets were added, along with PMSF to give a final concentration of 1 mM. The resuspended cells
were then sonicated using a Misonix S-3000 with six cycles of 20-sec pulsing (power setting = 5), followed by a 40-sec rest period.
Cellular debris was removed by centrifugation at 13,000 rpm using a JA-20 rotor in a Beckman L8-M Ultracentrifuge for 30 min at 4 C.
The supernatant was loaded onto immobilized Ni-NTA beads (1 mL bed volume). The column was washed with 40 mL wash buffer
(20 mM sodium phosphate, 500 mM sodium chloride, 10 mM imidazole, pH 7.4) and the protein was subsequently eluted with
5 separate additions of 2 mL elution buffer (20 mM sodium phosphate, 500 mM sodium chloride, 250 mM imidazole, pH 7.4). Elutions
containing tau protein were pooled and subsequently concentrated, and buffer exchanged into PBS pH 7.4 using Amicon Ultra-15
centrifugal filters (10,000 NMWL).
In preparations where significant tau proteolytic breakdown fragments were observed in the Ni-NTA elution fractions, the concen-
trated and buffer-exchanged sample was subsequently loaded onto a Superdex 200 10/300 GL (GE Healthcare) column using a
500 mL capillary injection loop. Protein was eluted from the column with PBS (pH 7.4) buffer at a flow rate of 0.25 mL/min with fraction
collection every 3 min (0.75 mL/fraction). Fractions containing intact tau protein with few or no proteolytic fragments were pooled and
concentrated as described above.

Crosslinking
Samples were crosslinked by the addition of CBDPS-H8/D8 (Creative Molecules Inc.), DSA-12C6/13C6 (Creative Molecules Inc.),
DSG-H6/D6 (Creative Molecules Inc.), or SDA (ThermoFisher Scientific) to final concentrations ranging from 0.1-0.2 mM (CBDPS,
DSA, and DSG) or 5 mM (SDA), at room temperature for 15 min. Reactions were quenched by adding ammonium bicarbonate to
a final concentration of 10 mM.
SDA crosslinking reaction mixtures were incubated for 10 minutes in the dark to allow the NHS-ester reaction to take place, fol-
lowed by 10 minutes of UV irradiation under a 25W UV lamp (Model UVGL-58 Mineralight lamp, UVG) with a 366-nm wavelength filter.
SDA reaction mixtures were quenched with 10 mM ammonia bicarbonate. A portion of each crosslinking reaction mixture was
checked by SDS-PAGE gel to see the extent of potential intermolecular crosslinked products. Aliquots were subsequently run on
SDS-PAGE and monomeric tau bands were exised and in-gel digested with trypsin (Shevchenko et al., 2006). The recovered protein
digests were lyophilized to dryness and stored at -80 C until analysis by LC-MS, at which time they were resuspended in 0.1% formic
acid (FA) containing 10 mM TCEP prior to autosampler loading.

LC-MS/MS Analysis
Mass spectrometric analysis was then performed using a nano-HPLC system (Easy-nLC II, ThermoFisher Scientific), coupled to the
ESI-source of an LTQ Orbitrap Velos or Fusion (ThermoFisher Scientific), using conditions described previously (Brodie et al., 2017).
Briefly, samples were injected onto a 100 mm ID, 360 mm OD trap column packed in-house with Magic C18AQ 100 E, 5-mm pore size
(Bruker-Michrom, Auburn, CA), and desalted by washing with Solvent A (2% acetonitrile:98% water, both containing 0.1% FA).
Peptides were separated with a 60-min gradient as follows: (0–60 min: 4–40% solvent B (90% acetonitrile, 10% water, 0.1% FA,
60–62 min: 40–80% B, 62–70 min: 80% B), on a 75-mm ID, 360-mm OD analytical column packed with Magic C18AQ 100 Å, 5-mm
pore size (prepared in-house), with IntegraFrit (New Objective Inc., Woburn, MA) and equilibrated with solvent A. MS data were
acquired using a data-dependent method. The data-dependent acquisition also utilized dynamic exclusion, with an exclusion win-
dow of 10 ppm and exclusion duration of 60 seconds. MS and MS/MS events used 60000- and 30000-resolution FTMS scans,
respectively, with a scan range of m/z 400-2000 in the MS scan. For MS/MS, the CID collision energy was set to 35%. Crosslinking
data were analyzed using Kojak (Hoopmann et al., 2015) and Percolator (Ka €ll et al., 2007) using filtering criteria of Percolator
q-value <= 0.01, score >= 1.675, population type = "intra", ppm error +/- 2.5.

Surface Modification
Chemical surface-modification data in form of photoreactive dead-ends were extracted from SDA crosslinking data sets. Amino acid
residues modified with the photoreactive moiety of the crosslinker, and the hydrolyzed amino-reactive moiety of the crosslinker (i.e., a
mass addition of 100.05 Da), were considered.

Discrete Molecular Dynamics Modeling

CL-DMD simulations were performed according to the protocol described in our previous work (Brodie et al., 2017). Briefly, in order to
incorporate the experimental data for inter-residue distances between specific atoms into the DMD simulations (Proctor et al., 2011),
we introduce a series of well-shape potentials that energetically penalize atoms whose interatomic distance do not satisfy the
experimentally-determined inter-atom proximity constraints. The widths of these potentials are determined by the crosslinker spacer
length and the side-chain flexibility (Brodie et al., 2017). Starting from a completely unfolded structure of the tau molecule, we
performed an all-atom Replica Exchange (REX) (Okamoto, 2004) simulation of the protein where 24 replicas with temperatures
equally distributed in the range from 0.375 to 0.605 kcal/(mol kB), are run for 6 x 106 DMD time-steps. We discarded the first
2 x 106 time-steps of system equilibration during the data analysis. In addition, we ranked all of the structures among all of the

Structure 27, 1–6.e1–e4, November 5, 2019 e3

 Please cite this article in press as: Popov et al., Insight into the Structure of the ‘‘Unstructured’’ Tau Protein, Structure (2019), https://doi.org/10.1016/
j.str.2019.09.003

simulation trajectories, and selected the ones with the lowest 10% of the energies, as determined by the DMD Medusa force field
function (Yin et al., 2008). These structures were then clustered using the GROMACS distance-based algorithm (Pronk et al.,
2013). The centroids of the most populated clusters were selected as our representative model of the tau protein structure. Because
there is no distinct lowest-energy state for the protein (because tau is intrinsically disordered), picking just one of these states would
potentially introduce a bias to our structure predictions. Thus, we present them all of these low energy states as our predicted models
of the tau conformational ensemble.

Unconstrained Molecular Dynamics Modeling

Unconstrained MD simulations were performed in the GROMACS 2018 package (Hess et al., 2008; Pronk et al., 2013) using
CHARMM36 force field (Best et al., 2012). Each structure was solvated using the TIP3P water model and neutralized with sodium
and chloride ions. Particle mesh Ewald (PME) was used for long-range electrostatic interactions with a 10A cutoff for non-bonded
interactions. Systems were initially equilibrated using a NVT thermostat and were later subjected to 4 heating-cooling cycles during
a 4-ns simulated annealing procedure. During these cycles the NVT equilibrated system (both the protein and the solvent) were
linearly heated from 300K to 325K for 0.5 ns and then cooled back to 300K during next 0.5 ns. This procedure was repeated 4 times.
After that, the systems were further equilibrated using NPT thermostats. Next, regular MD simulations were performed at a constant
pressure and temperature of 1 atm and 298 K for additional 50 ns per replicate. After the simulations were completed, we performed
clustering on the trajectories and selected a representative structure as a centroid of the most populated cluster.

QUANTIFICATION AND STATISTICAL ANALYSIS

Crosslinked peptide assignments were accepted with Percolator q-value <= 0.01, score >= 1.675.

DATA AND CODE AVAILABILITY

Structures of the centroids of the main clusters were deposited to PDB-Dev, and the accession code for our entry is PDBDEV_
00000033.
The CL-DMD software is available upon request from NVD (dokh@psu.edu). The mass spectrometry proteomics data have been
deposited to the ProteomeXchange Consortium via the PRIDE partner repository (Perez-Riverol et al., 2019) with the dataset
identifier PXD015044.

e4 Structure 27, 1–6.e1–e4, November 5, 2019