RESEARCH ARTICLE

Redox Proteomics www.proteomics-journal.com

Large-Scale Analysis of Redox-Sensitive Conditionally

Disordered Protein Regions Reveals Their Widespread
Nature and Key Roles in High-Level Eukaryotic Processes
Gábor Erdős, Bálint Mészáros, Dana Reichmann, and Zsuzsanna Dosztányi*

functions. For example, enzymes typi-

Recently developed quantitative redox proteomic studies enable the direct cally need to adopt a well-defined con-
identification of redox-sensing cysteine residues that regulate the functional formation that ensures the correct ori-
behavior of target proteins in response to changing levels of reactive oxygen entation of their active site residues, as
species. At the molecular level, redox regulation can directly modify the active required for optimal catalytic activity. In
contrast, intrinsically disordered proteins
sites of enzymes, although a growing number of examples indicate the and protein regions (IDPs/IDRs) lack a
importance of an additional underlying mechanism that involves conditionally well-defined structure in isolation, and
disordered proteins. These proteins alter their functional behavior by are best characterized by an ensemble of
undergoing a disorder-to-order transition in response to changing redox rapidly interconverting conformations.[1]
conditions. However, the extent to which this mechanism is used in various The specific structural properties of IDPs
enable them to carry out a different set of
proteomes is currently unknown. Here, a recently developed sequence-based
functions, which could not be fulfilled by
prediction tool incorporated into the IUPred2A web server is used to estimate relatively rigid globular domains.[2,3] As
redox-sensitive conditionally disordered regions at a large scale. It is shown a result, the function of many IDPs di-
that redox-sensitive conditional disorder is fairly widespread in various rectly originates from their high intrinsic
proteomes and that its presence strongly correlates with the expansion of flexibility, enabling them to act as link-
specific domains in multicellular organisms that largely rely on extra stability ers or spacers.[3,4] However, there is an
emerging understanding that the struc-
provided by disulfide bonds or zinc ion binding. The analyses of yeast redox tural behavior of IDPs is often context-
proteomes and human disease data further underlie the significance of this dependent: a disorder-to-order or order-
phenomenon in the regulation of a wide range of biological processes, as well to-disorder transition can be induced as
as its biomedical importance. a result of binding to specific macro-
molecular partners, undergoing post-
translational modifications, or changes
in environmental factors such as pH and
1. Introduction temperature.[5] Recently, redox conditions have emerged as an-
other important factor that can regulate conditional disorder.[6]
The key to the extraordinary versatility of proteins lies in their
The growing number of examples of redox-regulated condition-
specific structural properties that precisely suit their individual
ally disordered proteins indicates that this is an important mech-
anism triggered in response to various forms of oxidative stress
G. Erdős, Dr. B. Mészáros, Dr. Z. Dosztányi or to naturally occurring changes in redox potential.[6,7]
MTA-ELTE Lendület Bioinformatics Research Group Redox regulation is essential for the survival of all organisms,
Department of Biochemistry as cells constantly encounter the transient accumulation of re-
Eötvös Loránd University
active oxygen species (ROS). ROS may be generated exogenously
Budapest H-1117, Hungary
E-mail: [email protected] or endogenously due to metabolic activity or inflammation, while
Dr. B. Mészáros subsequent oxidative damage can cause widespread protein un-
Structural and Computational Biology Unit folding and aggregation.[8,9] It may also contribute to the patho-
European Molecular Biology Laboratory genesis of many different diseases, as well as to degenerative pro-
Heidelberg 69117, Germany cesses associated with aging.[10,11] However, ROS can also serve
Dr. D. Reichmann as cellular signaling molecules and regulate various biological
Department of Biological Chemistry
The Alexander Silberman Institute of Life Sciences processes such as the immune response, cell proliferation, de-
Safra Campus Givat Ram velopment, and more.[12,14] The specific effects of ROS are cap-
The Hebrew University of Jerusalem tured in large part through the covalent and reversible modifi-
Jerusalem 91904, Israel cations of specific cysteine residues, which can in turn modify
the structural or functional properties of the redox-sensitive pro-
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/pmic.201800070 teins. Such posttranslational modifications are tightly regulated
by changes in levels of diverse physiological oxidants, and are
DOI: 10.1002/pmic.201800070

Proteomics 2019, 1800070 1800070 (1 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

generally faster than expression of regulatory proteins. Over the

last decade, numerous examples of such thiol-switch proteins Significance Statement
have been uncovered, spanning different functions and regula- In this work, we analyzed the occurrence of redox-regulated
tion modes.[6,7,15] The functions of these proteins were found to conditionally disordered proteins at the level of proteomes
be regulated by oxidative modifications of a specific cysteine thiol using a novel computational tool. An increasing number
forming either sulfenic (e.g., in 1-Cys peroxiredoxin) or sulfe- of examples indicates that an important mechanism of re-
namide (e.g., in tyrosine phosphatase 1B, leading to regulation dox regulation involves conditionally disordered proteins,
of the Ras signaling pathway). Another more studied mode of ac- proteins that alter their functional behavior by undergoing a
tivation is through disulfide bond formation by redox-sensitive disorder-to-order transition upon changing oxidative condi-
cysteine residues, typically located in close proximity of each tions. However, the extent to which this mechanism is used
other. These disulfide bridges can either promote disorder-to- in various proteomes is currently unknown. Here, we showed
order transitions (e.g., chloroplast regulator of the calvin cycle, that redox-sensitive conditional disorder is fairly widespread
CP12[16] , and cell growth modulator, granulin Grn-3[17] ) or vice in various proteomes and also occurs among proteins identi-
versa (e.g., in the bacterial protein chaperone Hsp33[18] and mam- fied by proteomics studies as redox regulated. The presence of
malian copper chaperone Cox17[19] ) in response to shifts in the such regions strongly correlates with the expansion of specific
cellular redox status.[6] domains at various points of evolution. Redox-regulated condi-
The growing interest in redox regulation has motivated the de- tionally disordered proteins not only play key roles in high-level
velopment of various proteomics approaches that can be used eukaryotic processes but their mutations can contribute to
to explore the landscape of thiol redox modifications. These several diseases. The presented analysis highlights the sig-
methods usually involve differential modification of the thiol nificance of the interplay between redox regulation, structural
groups by specific probes that label reduced thiolate groups. changes, and proteostasis related functions, and can direct
Since the oxidized cysteines do not interact with these probes, further experimental studies.
the typical workflow involves an initial modification of all re-
duced thiols in the cell proteome, followed by reduction and
then labeling with the second probe that differs in mass from
the first. Thus, reduced and oxidized cysteines can be differ- in the proximity of each other within structures of globular
entially labeled in a ratiometric manner, allowing for quantita- proteins, as compared to a background model. In this regard,
tive analysis of reversible oxidation modification of thiol groups. cysteine residues clearly stand out, as they often form covalent
One of these approaches is based on the differential labeling contacts with other cysteines, thus providing large stabilizing
by N-ethylmaleimide (NEM) and N-(6-(biotinamido)hexyl)-3 (2 - contributions either by forming disulfide bridges or coordinating
pyridyldithio)propionamide (biotin-HPDP), which enables the Zn2+ and other metal ions. Consequently, cysteine residues have
analysis mainly of the reversibly oxidized thiols.[20] In contrast, a strong order promoting character in IUPred. However, thiol
the OxICAT methodology is based on differential labeling of re- modifications of either paired or single cysteine residues may
duced and oxidized cysteines with isotopic light and heavy forms lead to a reversible order-to-disorder transformation, exposing
of the coded affinity tag (ICAT) reagent, derivatives of the iodoac- unfolded regions crucial for protein function.[6,7] This regulatory
etamide alkylation reagent.[21] The usage of two differential iso- effect of redox conditions on protein structure not only follows
topes provides quantification of the degree of oxidation among the presence of oxidation sensitive cysteine residues, but is also
thousands of specific cysteine residues within the analyzed pro- influenced by the presence of disorder-promoting residues in
teome in a highly precise manner. OxICAT has thus far been ap- the surrounding sequence. In such cases, cysteine residues
plied to proteomes (redoxomes) of diverse organisms under dif- may simply be considered small, polar residues akin to serine,
ferent physiological conditions.[22–25] located within a sequence region exhibiting distinct features.
Using redox proteomics studies, the cellular redox state has The prediction of redox-sensitive disordered regions is based
thus far been revealed to either directly or indirectly affect a wide on capturing this dual characteristic of cysteines. This novel
range of physiological processes. However, proteomics studies method opens new doors to explore redox-sensitive conditionally
neither provide information about the underlying mechanisms disordered segments in various proteomes.
of thiol modifications, nor to what degree the detected thiol Here, we apply our method to predict redox-sensitive disordered
modifications correspond to redox-regulated conditional disor- regions on a large scale. We analyze the occurrence of these
der. In order to explore this aspect of redox regulation, we took segments in known structures, and connect them to structural
advantage of a recently introduced new feature of the IUPred and functional features. In order to explore evolutionary trends
method (IUPred2A) that enables the sequence based prediction of redox-sensitive conditionally disordered regions, we estimate
of regions that are likely to undergo disorder-to-order transitions their abundance at the proteome level and analyze their enrich-
upon redox changes.[26] IUPred is a robust sequence-based ment in terms of domains and biological processes. By taking ad-
prediction method for protein disorder that uses an energy vantage of recent MS-based techniques applied to the exploration
estimation method.[27] The strength of the method originates of redox-sensitive thiols, we can highlight proteins that are likely
from its ability to capture the basic biophysical properties of to use disorder-to-order transition to respond to redox changes in
disordered segments: their inability to form enough stabilizing biological settings. We have used experimentally derived datasets
interactions to adopt a well-defined globular structure. The core generated by two different redox probes: either ICAT (used
of the method is a statistical potential that assigns more favorable in the OxICAT redox proteomic method)[21] or biotin-HPDP.[20]
scores to amino acid pairs that are observed more frequently Finally, by analyzing disease data, we collected examples of

Proteomics 2019, 1800070 1800070 (2 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

genetic mutations that alter the behavior of putative redox- redox-sensitive conditionally disordered proteins and fredox , are
sensitive regions. listed in the Supporting Information.
The set of human proteins containing predicted redox-
sensitive conditionally disordered protein regions (RSCDP) was
2. Experimental Section determined as described for all proteomes. The list of 5643 pro-
teins together with the 10 108 identified regions is listed in the
For the prediction of redox-regulated conditionally disordered Supporting Information.
segments, a slightly modified version of the IUPred2A method Redox-sensitive conditionally disordered domains were iden-
was used. The basic principle is the same as published.[26] Two tified by assessing overlapping regions predicted by IUPred2A,
prediction profiles are generated with the default settings of the and domains predicted by Pfam[30] (release 31.0). A Pfam domain
IUPred method, one for the native sequence corresponding to instance was considered to be redox-sensitive conditionally dis-
the state achieved through cysteine stabilization (redox-plus) and ordered if an IUPred2A region covers at least five of its residues
one for a modified sequence with cysteines residues mutated to and at least 20% of its total length. The number of thus identified
serine, corresponding to the state without cysteine stabilization redox-sensitive domains is listed in the Supporting Information
(redox-minus). To initiate a redox-sensitive conditionally disor- for the S. cerevisiae, D. melanogaster, and H. sapiens proteomes,
dered region, the difference between the two profiles has to be containing 6049, 13 775, and 21 050 proteins, respectively. In
larger than 0.3, while the minus profile is below 0.5 and the each case, the reference proteomes downloaded from UniProt
plus profile is above this threshold at a given position. This re- were used.
gion is then extended until the difference between the two pro- Gene Ontology (GO) annotations for human proteins were
files become lower than 0.15, or the redox plus profile fell be- taken from the 24/05/2018 version of the EBI GOA database.[31]
low 0.35. Two adjacent regions are combined if they are closer Only those biological process terms were considered that are at-
than ten amino acids. Two minor modifications were introduced tached to at least one protein with at least one predicted redox-
compared to the original web server implementation, aimed at sensitive structural switch. In order to enable the analysis of fairly
decreasing potential overprediction of such regions. First, a pre- generic and high level processes, only those terms were kept that
dicted redox-sensitive conditionally disordered region must have were reachable from the root term (biological process) with at
at least three residues that meets the opening criterion. Sec- most four steps following only ‘is a’ or ‘part of’ relations. These
ond, the minimal length of the predicted region must be at least terms were further filtered for ancestry, and if two terms were
15 residues. The current version of the program is available at: in ancestor/child relationship, only the child term was kept. The
http://iupred2a.elte.hu resulting set of GO terms was manually split into four groups
For the analyses of Protein Data Bank (PDB) structures, a non- based on the level of biological process they describe. This yielded
redundant dataset of protein structures was downloaded from the the GO molecular slim, containing 61 terms describing basic
PISCES web server, filtered at 40% sequence similarity, resolu- molecular processes; the GO network slim, containing 77 terms
tion of at least 2 Å, and maximum R factor of 0.25.[28] The dataset describing pathway/network level processes; GO cell slim, with
contains 14 150 protein chains. Five chains were omitted for tech- 103 terms describing cellular processes; and GO organism slim,
nical reasons, and for the remaining sequences the redox regions with 183 terms describing organism level biological processes.
were predicted with the customized IUPred2A program. Disul- The enrichment of each term among redox-sensitive condition-
fide bonds were identified from the structure using the distance ally disordered proteins was assessed by comparing the occur-
criterion of 2.3 Å between the sulfur atoms of two cysteines. For rence of the term in the RSCDP set and the human proteome.
the identification of cysteine clusters, all potential combinations 1000 sets of 5643 proteins were taken from the human proteome
of four non-disulfide bonded cysteine and histidine residues were randomly, and were evaluated for term occurrences, yielding an
considered within each structure. For the four selected residues, average and standard error for expected occurrence. The enrich-
the approximate location of the metal center was calculated as ment of each term in RSCDP was expressed as the difference be-
the center of mass of the positions corresponding to the coordi- tween the observed and expected occurrences in standard error
nates of SG atoms for cysteines and ND1 atoms for histidines. A deviations. The Supporting Information shows the enrichment
cysteine cluster was identified when the distance of each of the values for all terms in all four slims.
considered atoms from this center position was smaller than 4.2 Missense variants annotated in human UniProtKB/Swiss-
Å. These criteria were selected to maximize the agreement with Prot entries were downloaded from the Humsavar web-
the provided PDB annotations. site (https://www.uniprot.org/docs/humsavar version 07/2018).
Full proteomes were downloaded from the 23/08/2018 version Only those mutations were considered that contained “Disease”
of UniProt reference proteomes.[29] The customized IUPred2A annotation as type of the variant and involved a cysteine either
was run on all protein sequences from all proteomes, using the in the original or in the mutated form (3955 mutations). Muta-
redox-sensitive option. A protein was considered to contain a tions located within redox-sensitive conditionally disordered re-
redox-sensitive conditionally disordered region if IUPred2A re- gions were identified based on an overlap with predicted regions
turned at least one region. The fredox value of a proteome was cal- using the original sequence in cases when a cysteine residue was
culated as the number of proteins with predicted regions over altered, or using the mutated sequence in case the mutation in-
the total number of proteins in the proteome. In total, the an- troduced a cysteine. The identified mutations are shown in the
alyzed data consisted of 192 viral, 436 archaeal, 608 bacterial, supporting information. To concentrate on cases that are likely to
and 167 eukaryotic proteomes, including multiple plant organ- be directly associated with alteration of a redox-regulated condi-
isms. The proteomes, together with the number of predicted tionally disordered region, proteins were sorted according to the

Proteomics 2019, 1800070 1800070 (3 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

ratio of mutations within predicted regions versus all mutations association with the Mia40 protein. In agreement with experi-
within the corresponding gene. mental data, the whole protein—apart from its signal sequence—
is predicted to undergo a disorder-to-order transition upon
changes of redox potential (Figure 1).
3. Results The oxidation driven order-to-disorder transition in the case of
Hsp33 and the disorder-to-order transition of COX17 produce the
3.1. Prediction of Redox-Sensitive Structural Switches in Proteins active form of the proteins, however, such transitions can also be
corresponding to Known Examples part of more complex regulatory mechanisms. For example, the
cytoplasmic tail of T cell coreceptors CD4 and CD8 associate with
Redox-dependent disorder-to-order transitions have been de- the N-terminus of the Src-family tyrosine kinase Lck. These in-
scribed in detail in only a few cases.[6,7] For this reason, rigor- teractions are critical for T cell development and activation. The
ous benchmarking on a large number of positive and negative interacting tail regions of both CD4 and Lck were shown to be in-
data is not possible for the prediction of redox-sensitive struc- trinsically disordered in isolation, yet assemble to a form a zinc
tural transitions. Instead, IUPred2A was established and is fur- clasp structure.[39] Interestingly, the interaction is also regulated
ther customized in this study, to best describe the available exper- at the level of alternative splicing.[40] The interaction prevents the
imentally verified examples. The prediction is based on the gen- internalization and degradation of CD4 by masking the dileucine
eration of two prediction profiles, one using the original amino motif required the clathrin-mediated endocytosis.[39] While there
acid sequence (redox-plus, assuming cysteine stabilization) and is no direct evidence for redox regulation of this complex, activa-
one with changing each cysteine to serine (redox-minus, corre- tion and proliferation of CD4+ T cells is associated with the in-
sponding to the lack of cysteine stabilization). While the predic- tracellular increase of reactive oxygen species and the release of
tion scores would differ for any sequence containing cysteines, zinc.[41] This suggests that the Lck-CD4 complex could be consid-
in most cases, no changes in the overall order-disorder tendency ered as an additional example of the redox-regulated conditional
would be observed. Our hypothesis was that in the cases of se- disorder (Figure 1).
quences that undergo disorder-to-order transition upon changes
in redox conditions, a pronounced shift could be observed with
the redox plus profile predicted as ordered, and the redox minus
profile predicted as disordered. Redox-sensitive regions involved 3.2. Redox-Sensitive Conditionally Disordered Regions in the
in this type of transition are predicted based on the observed di- PDB
vergence of the two profiles (see Experimental Section).
The performance of our prediction method is illustrated The number of fully characterized redox-induced fold-
through three interesting cases. One of the best characterized ing/unfolding examples is rather limited. However, the
examples is the Hsp33 heat shock protein from Escherichia coli. developed method can be further characterized and refined
This protein chaperone forms a well-defined structure under re- using the vast amount of information encoded in protein
ducing conditions, with the N-terminal domain in close con- structures in the PDB containing either oxidized or reduced
tact with the adjacent linker region and the Zn2+ binding C- forms of cysteines. To this end, we collected structures from the
terminal domain.[32] However, substrate binding is inaccessible PDB using a filtering of 40% sequence identity. The predictions
in this conformation, and consequently, the protein lacks chaper- were run using the sequence from the corresponding PDB file.
one activity.[18] Oxidative unfolding (oxidation coupled with mild The presence of disulfide bonds was calculated from the atomic
protein destabilizing conditions) causes the C-terminal region to coordinates. However, cysteine residues that are involved in
release the Zn2+ ions, inducing the unfolding of the C-terminal coordinating metal ions are also often located in close proximity.
domain together with the adjacent linker region. This transi- Such clusters were also identified based on the coordinates (see
tion exposes the substrate binding sites, that makes the protein Experimental Section). The statistics for each PDB chain in the
active.[33,34] The critical region for redox-sensing is located be- non-redundant dataset is given in the Supporting Information.
tween residues 230 and 266, which contains the four highly con- Redox-sensitive regions were predicted for 711 out of 14 145
served cysteine residues that coordinate the Zn2+ ion under re- structures, representing only 5.03% of cases (Figure 2A). Most of
ducing conditions, but form short-range disulfide bonds under the redox-sensitive regions corresponded to structures stabilized
oxidative conditions. The large part of the experimentally deter- by disulfide bridges (417, 2.95%), or cysteine clusters, coordinat-
mined redox-sensitive region is correctly identified by IUPred2A ing either Zn2+ , Cu2+ , or Fe2+ clusters (214, 1.51%). However,
(Figure 1).[35,36] 1566 structures with disulfide bonds and 344 structures with
Another prototypical example of redox-regulated disorder-to- cysteine clusters did not contain predicted redox-sensitive
order transitions is the human COX17, a ࣈ60-residue copper structural switches. In general, structures with redox-sensitive
chaperone that is one of the 13 subunits of the mammalian cy- regions had a higher number of disulfide bonds or cysteine
tochrome c oxidase complex. This protein is disordered under the clusters (Figure 2B,C).
reducing conditions of the cytosol, in which it is synthesized, and While disulfide bridges usually form under oxidative condi-
its disordered nature is essential for its diffusion across the mito- tions, metal binding is more common in reducing environments.
chondrial outer membrane.[19,37,38] However, upon its entry into Consequently, these structural features are usually complemen-
the oxidizing environment of the intermembrane space of the tary, and there are only a few structures that contained both
mitochondria, COX17 undergoes a disorder-to-order transition, cysteine clusters and disulfide bonds. While in certain cases
and its folding proceeds by sequential disulfide formation and the disulfide bond formation could be a crystallization artifact

Proteomics 2019, 1800070 1800070 (4 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

Figure 1. Known examples of redox-change induced structural transitions. Structural figures: cysteines are shown in blue, metal ions are marked as
green spheres. The di-leucine degradation motif of CD4 is shown in red. Green ticks and red crosses mark the active and inactive states of the pro-
teins/complexes. The prediction output is taken from the IUPred2A web server, with purple shadings marking the predicted redox-sensitive regions, and
blue ovals showing the positions of the redox-sensitive Cys residues. UniProt accessions used: P0A6Y5 (Hsp33), Q12287 (COX17), P01730 (CD4), and
P06239 (Lck). PDB IDs used: Hsp33|reduced (1xjh), COX17|oxidized (1z2g), and CD4:Lck|reduced (1q68). Structures of Hsp33, CD4, and Lck feature
only the redox-sensitive regions.

(e.g., 2g45), in many cases, the complementary nature of metal structures adds strong credibility to IUPred2A redox-sensitive
binding and disulfide bond formation can ensure a stable predictions.
structure under a variety of conditions. As a counterpoint to
structures with both of the stabilizing features, a relatively
small number of predicted proteins (88, 0.62%) did not have
either. The majority of these structures describe heme binding 3.3. Redox-Sensitive Structural Switching Correlates with
proteins, which rely on heme groups coordinated partially Evolutionary Complexity
by cysteines for their stability, and might in fact be involved
in redox regulation.[42] However, disorder-to-order transitions IUPred2A is able to predict redox-sensitive protein regions based
are not necessarily coupled to redox regulation. In general, on sequences alone. This opens up the means of studying the
examples without the capacity to form either disulfide bonds or large-scale distribution of such regions across several proteomes
cysteine clusters are treated as likely false positives. However, (see Experimental Section), to gain evolutionary insights into the
the number of such cases in PDB structures represent less emergence of redox-sensing structural switches. We assessed the
than 1% of all structures in our dataset (Figure 2A). Overall, the fraction of proteins with redox-sensitive regions with a poten-
strong correlation of redox-sensitive conditionally disordered tial structural plasticity in reference proteomes from all three do-
regions with strongly stabilizing structural features in known mains of life, as well as viruses (Figure 3A).

Proteomics 2019, 1800070 1800070 (5 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

lowest fredox values, with over 90% of species having fredox < 7%.
In general, for unicellular organisms, fredox is below 10%, includ-
ing almost all archaea and bacteria, together with some eukary-
otes. In contrast, multicellular organisms are characterized by
markedly higher fredox values, with some highly complex organ-
isms reaching close to 30%. The steady increase of fredox with the
increase of complexity is demonstrated by the comparison of the
Saccharomyces cerevisiae (fredox = 7.2%), Drosophila melanogaster
(fredox = 19.6%), and human proteomes (fredox = 27.2%). Over-
all, the separation between unicellular and multicellular organ-
isms is much pronounced in the case of fredox values, indicating
that fredox values (as opposed to genome or proteome sizes) bet-
ter reflect the apparent complexity of organisms. The content of
predicted disordered regions (based on IUPred2A) and predicted
disordered binding regions (based on ANCHOR2) also increases
with the complexity of organisms (Figures S1 and S2, Supporting
Information). Most proteomes have a higher percentage of pro-
teins with disordered regions than with redox-sensitive intrinsi-
cally disordered regions. These differences are even more appar-
ent at the per residue level. Taking the human proteome as an
example, the percentage of disordered residues can be as high as
25%, but less than 5% of the residues are predicted to be located
in redox-sensitive disordered regions (see Supporting Informa-
tion). Consequently, the content of disordered regions and dis-
ordered binding regions show only a weak correlation with the
fraction of predicted redox-sensitive intrinsically disordered pro-
teins (Figures S3 and S4, Supporting Information).
For complex organisms, the derivation of even the proteome
size from the genome sequence is non-trivial.[43] This is further
modulated by the presence of various isoforms; therefore, in the-
ory, fredox values could be highly dependent on the exact defini-
tion of the proteome used for calculations. However, testing for
this effect showed that in reality the calculated fredox values are
strikingly stable with respect to such variations in proteome defi-
nitions. The minimal proteome of D. melanogaster from UniProt
contains 3496 proteins, while the proteome featuring all isoforms
contains 13 775 proteins. Despite this nearly fourfold difference,
both proteomes yield almost identical fredox values of 19.6% and
20.8%, respectively. This supports the robustness of fredox as a
marker for organism complexity, with the largest breakpoint cor-
responding to the advent of multicellularity.
Apart from multicellular eukaryotes, several viruses also ex-
Figure 2. Predicted redox-sensitive regions in the PDB. A) The overlap be-
tween proteins with predicted redox-sensitive regions, and proteins with hibit high fredox values. In these cases, fredox values correlate with
disulfide bonds/metal ion Cys clusters in the PDB. The size of circle does the complexity of the host, with most high fredox value viruses
not scale exactly with the size of datasets. The distribution of structures targeting complex eukaryotes, typically vertebrates. This corre-
with respect to disulfide bridge (B) and metal ion coordinating Cys clus- lation between host complexity and fredox value is also apparent
ters (C) for structures with and without predicted redox-sensitive regions. for several unicellular eukaryotic pathogens. Both T. congolense
Dashed lines mark the average values for all distributions.
and P. falciparum (the major causes of the sleeping sickness and
malaria, respectively) are unicellular eukaryotes, and such their
fredox values are expected to be below 10%, similarly to other
Genome size generally reflects the average complexity of the organisms with comparable complexity and small proteome
domains of life. While the differences between specific organ- size. However, their fredox values rather resemble those of their
isms are obscured by the C-value paradox, generally viruses hosts, with fredox = 24.4% and 17.9%, respectively (Figure 3A).
have the smallest genomes, followed by archaea, bacteria, and While the fraction of proteins incorporating redox-sensitive in-
eukaryotes. This hierarchy is also reflected in the proteome trinsically disordered regions increases with organism complex-
sizes, albeit with large overlaps between various domains of life. ity, the typical length of such regions is strikingly constant across
However, the fraction of predicted redox-sensitive intrinsically various domains of life and viruses (Figure 3B). The typical re-
disordered proteins (termed fredox , see Experimental Section) gion length is ࣈ22 residues, and while eukaryotes have a slightly
shows a distinctively different distribution. Bacteria feature the higher average region length, this is primarily a result of the high

Proteomics 2019, 1800070 1800070 (6 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

Figure 3. Proteome-wide distribution of redox-sensitive protein regions. A) The fraction of proteins containing redox-sensitive regions as a function of
proteome size (on a logarithmic scale). Red boxes mark pathogens, with the host organisms following the name. B) Length distribution of redox-sensitive
regions across various domains of life and viruses. C) The number of predicted redox-sensitive metal ion coordinating (red) and disulfide bonded (blue)
conditionally disordered structural units (left vertical axis). Proteome sizes are shown in grey bars for reference (right vertical axis). D) The frequency
of various Pfam domains among the predicted redox-sensitive structural switches (see Supporting Information for full lists). Red boxes mark metal ion
coordinating domains, and blue boxes mark disulfide bonded domains. Only domains with a relative frequency of 1.25% or higher are shown in separate
boxes, other domain types are merged and shown as “other.”

Proteomics 2019, 1800070 1800070 (7 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

apparent length of tandem arrays of redox-sensing regions. This other hand, the other two datasets were obtained using the
indicates that the basic molecular mechanisms behind redox- OxICAT method, which analyzes all cysteines, and provides
sensing structural switches are probably uniform across all life, degree of oxidation of the identified cysteines (reduced and
and the emergence of complexity requires a larger number of pro- oxidized in the protein lysate), focusing on a wider range of
teins utilizing such mechanisms. cysteine containing peptides. The second study applied quan-
In order to directly assess the structural units involved in titative redox proteomics based on the OxICAT approach[44]
redox-sensitive order-to-disorder transitions, we also analyzed and identified 41 proteins that underwent substantial thiol
the domains that overlap with the predicted regions in the three modifications as a result of H2 O2 treatment. Using similar thiol
model organisms highlighted in Figure 3A (see Figure 3C). In labeling but different MS workflow, the third study identified 47
the yeast proteome, nearly all redox-sensitive structural switches unique proteins with redox-active thiols.[45] Strikingly, there was
are metal ion coordinating domains. The majority of these do- very limited overlap between these datasets. Only five proteins
mains coordinate zinc ions, belonging to a handful of zinc finger were common between the first two studies (Trp5, Elf1, Fba1,
domain types (Figure 3D). The transition into the realm of mul- Aro4, and Ahp1), and only a single protein (Rpl37B) was shared
ticellular organisms brings about a drastic increase in the rela- between the second and third studies, with no overlap between
tive number of zinc ion coordinating structural switches. While the first and third studies. This small overlap among the three
the ratio of proteome sizes for yeast versus fruit fly is only 2.3 studies seems to indicate limitations in terms of the currently
(6049 vs 13 775), the ratio of the number of identified metal applied techniques’ sensitivity, as well as technical and biological
ion coordinating domains is about 4.5. Analysis of the fruit fly variabilities.
proteome also demonstrates the emerging dominance of C2H2- The structural or functional consequences of cysteine modi-
type zinc fingers. In addition, the fruit fly proteome also shows fications identified in these experiments can be heterogeneous,
a striking increase in the number of disulfide-bound structural and not all of them are expected to induce a transition in the struc-
switches. In contrast to metal ion binding domains, however, the tural state of the affected region. Nevertheless, in several cases,
identified disulfide bonded domains are more evenly distributed the identified proteins with modified cysteines overlapped with
across several types of domains. Considering the human pro- predicted conditionally disordered regions. As these predictions
teome, the abundance of both types of structural switches are suggest a specific mechanism underlying the redox-sensitivity,
even higher. However, this increase is more pronounced for the we took a closer look at these examples (Table 1). In one ex-
metal ion binding domains, and is largely attributable to the dras- ample, the Gas1 protein is anchored to the plasma membrane
tic increase in the number of C2H2 type zinc finger domains. and to the cell wall. In addition to the catalytic glucanosyltrans-
In comparison, extracellular, disulfide bonded domains remain ferase domain, it also contains a cysteine rich domain. The redox-
fairly diversified across several domain types (see Figure 3D and sensitivity of these cysteines is not easy to explain, as they are
Supporting Information). required for the normal folding and stability of the protein.[46]
Judging by the sheer numbers of occurrence, the most abun- An interesting case corresponds to Ydj1, a yeast homolog of the
dant domain of interest in humans is the C2H2 zinc finger. chaperone DnaJ, a zinc containing co-chaperone, involved in mi-
This structural unit is present in all three studied proteomes in tochondrial protein import. The central cysteine-rich domain,
high numbers. Furthermore, the ratio of conditionally disordered which contains four repeats of the motif CxxCxGxG, is predicted
C2H2 domains compared to the total number of such domains as conditionally disordered. Its redox-sensitivity suggests that it
is surprisingly stable across the three organisms, with values of could be involved in oxidative-stress response, similarly to the
59.5%, 66.6%, and 53.9%, for yeast, fruit fly, and human, respec- Hsp33 chaperone. The validity of the prediction is supported by
tively. This indicates that the study of the redox-sensitivity of sim- the notion that redox regulation was established for the human
ple model organisms, such as yeast, may in some cases have homolog of DnaJ.[47]
implications in human physiology as well, given that two out of Several members of the ribosome were also shown to be redox-
the three most common redox-sensitive domains in yeast (C2H2 sensitive, further containing predicted conditionally disordered
and CCHC type zinc fingers) are also present in the human pro- regions. This group includes the mitochondrial Mrpl32, which
teome in high numbers (see Supporting Information). The iden- was shown to undergo redox-dependent regulation.[48] Members
tification and functional characterization of redox-sensing struc- of the cytosolic ribonucleoprotein complex were also included
tural switches can therefore serve as a guide to the understanding in this group (Rpl37A, Rpl37B, Rps29A, Rps29B, Rpl40A, and
of the more complex roles these protein regions play in human Rpl40B). In addition, Mak16 is involved in the biogenesis of the
regulation. ribosomal large subunit in the nucleolus. The common struc-
tural theme among these proteins is Zn2+ binding, typically in-
volving a specific zinc finger class, called treble clefs. Additional
proteins involved in translational or transcriptional processes are
3.4. Comparison with Redox Proteomics Datasets in Yeast also candidates for redox-regulated conditional disorder medi-
ated by their Zn2+ binding domains, such as Taf1 and Elf1. Taf1 is
We analyzed three datasets of S. cerevisiae proteins containing a zinc knuckle that plays a central role in TfIID promoter binding
modified cysteines identified using redox proteomics. In the and regulation of transcription initiation.[49] Elf1 is a conserved
first case, 27 proteins were identified harboring reversibly transcription elongation factor that contains a putative zinc
oxidized cysteines using biotin-HPDP-based redox binding domain with four conserved cysteines.[50] Another Zn2+
proteomics.[20] The biotin-HPDP-based method identifies binding protein, Gis2 interacts with the mRNA translation
only cysteines that were oxidized in the protein lysate. On the machinery and is a component of the stress induced Rnp

Proteomics 2019, 1800070 1800070 (8 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

Reference
granules.[51] Nrp1 also belongs to the group of Zn2+ finger pro-
teins, with a currently uncharacterized RNA binding function,
[20]

[44]

[45]

[45]

[45]

[45]

[45]

[45]

[45]

[45]

[45]

[45]

[45]

[45]

[45]
also localizing to stress granules. From a functional point of view,
the redox-sensitivity of these proteins can play a critical role in the
global attenuation of translation in response to oxidative stress.[45]

coordinating)
Further candidates for redox-regulated conditional disorder in-

Ribosomal L32p (Zn2+ coordinating)

Ribosomal L40e (Zn2+ coordinating)
Ribosomal L37e (Zn2+ coordinating)

Ribosomal L28e (Zn2+ coordinating)

Ribosomal S14 (Zn2+ coordinating)

clude other Zn2+ binding proteins, such as the DNA damage-

zf-CCHC 6 (Zn2+ coordinating)

responsive transcriptional repressor Rph1, pre-mRNA-splicing

zf-RanBP (Zn2+ coordinating)

zf-CCHC (Zn2+ coordinating)

zf-C2H2 (Zn2+ coordinating)

DnaJ C (Zn2+ coordinating)

coordinating)
factor Rds3, and DNA-directed RNA polymerases I, II, and III

Pet191 N (disulfide bond)

2+
Elf1 (Zn2+ coordinating)

DNA RNApol 7kD (Zn

subunit Rpabc4 (Table 1).
Affected Pfam domain

X8 (disulfide bond)

In contrast to the other examples, redox-sensitive Pet191 is

likely to gain structure under oxidative conditions when it en-

2+
ters the intermembrane space of the mitochondria that induces

PHF5 (Zn
disulfide-bond formation. This protein is a cytochrome c oxidase
assembly factor with a twin-Cx9 C motif, similar to that found in
Cox17. The disulfide bond formation is required for cell viability
and assembly of the cytochrome c oxidase complex, however, the
Predicted region

protein may be imported into the mitochondria independently of

the mitochondrial MIA40-ERV1 machinery.[52] This could explain
1031-1053

why this protein was found to be redox-sensitive, while some of

135-213

351-387
360-385

733-772
82-124

97-116
11-31

15-61
18-51

1-152

38-59
17-59

the other members of the COX family were not detected as such.
9-46

1-55
Table 1. Experimentally characterized redox-sensitive yeast proteins containing predicted conditionally disordered regions.

Redox cysteines

3.5. Functional Repertoire of Redox-Sensitive Structural Switches

1038,1041
19, 34, 37

25, 28, 52

in the Human Proteome

185, 188

104, 107
24, 39

31, 34
115

107
378
421

740
38

23
32

Yeast proteomics results highlight some key molecular mecha-

nisms in which redox-sensitive conditional disorder plays func-
tional roles, most notably in stress response, biogenesis, and the
DNA-directed RNA polymerases I, II, and III subunit Rpabc4

regulation of transcription and translation. These results indicate

DNA damage-responsive transcriptional repressor Rph1

that these structural switches preferentially contribute to the ac-

Ubiquitin-60S ribosomal protein L40 Rpl40A, Rpl40B

Transcription initiation factor TFIID subunit 1 TfIID

tuation of certain biological processes. While yeast proteomics

54S ribosomal protein L32, mitochondrial Mrpl32
Mitochondrial protein import protein MAS5, Ydj1

can highlight certain functions also present in higher order or-

40S ribosomal protein S29-B Rps29A, Rps29B

60S ribosomal protein L37-B Rpl37A, Rpl37B

ganisms, biological processes that are unique to higher order

multicellular organisms can only be uncovered via the direct anal-
Transcription elongation factor 1 Elf1
1,3-beta-glucanosyltransferase Gas1

ysis of the functional annotations of their proteomes.

The characteristic functional repertoire of human redox-
Pre-mRNA-splicing factor Rds3
Mitochondrial protein Pet191
Asparagine-rich protein Nrp1

sensitive conditionally disordered proteins was assessed using

GO term enrichment analysis (Figure 4). We identified terms
Zinc finger protein Gis2

that are significantly enriched in the set of predicted condition-

ally disordered proteins compared to the full human proteome
Protein Mak16
Protein name

(see Experimental Section). Relevant GO terms were divided into

four groups, termed slims, to separately highlight basic molec-
ular (GO molecular slim), network/pathway level (GO network
slim), cellular level (GO cell slim), and organism level biological
processes (GO organism slim).
Molecular level functions highlight that the dominant
RPS29A, RPS29B
RPL40A, RPL40B
RPL37A, RPL37B

processes uncovered by yeast proteomics are also heavily

Gene name

associated with redox-sensitive conditional disorder in the

MRPL32

PET191
MAK16

human proteome (Figure 4A). These processes are linked to the

RPC10
RPH1
NRP1
GAS1

RDS3
TAF1
YDJ1

ELF1

GIS2

regulation of transcription and translation, and are dominated

by intracellular proteins. Consequently, the corresponding GO
terms are clearly linked to metal ion coordinating regions. In
Uniprot accession

addition, these structural elements are also associated with

the biosynthesis of several organic compounds, processes that
are largely absent from the results of yeast proteomics. While
P0CH09

Q06835
Q02772
P22146
P25491
P41058

P51402
P25348
P10962
P36053
P46677
P53849
P32770
P39956
P40422

corresponding processes do exist in yeast, they involve only

a very limited number of proteins in contrast to the similar

Proteomics 2019, 1800070 1800070 (9 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

Figure 4. GO term enrichment of human redox-sensitive conditionally disordered proteins. Panels (A–D) show the 50 GO biological process terms that
are most enriched, compared to the full human proteome. Colors denote the dominance of either metal ion coordinating, or disulfide bonded structural
elements. Terms that are associated with both structural elements with equal frequency are shown in grey. A) Molecular processes, B) pathway/network
level processes, C) cellular processes, and D) organism level processes. For members of all four slims, together with exact numbers of occurrences and
enrichments, see Supporting Information.

function in humans (e.g., “nucleobase-containing compound chemotaxis, which mostly involves disulfide bonded extracellular
metabolic process” covering only seven yeast proteins[53] ). A proteins, and stem cell division and population maintenance,
separate class of GO molecular slim terms correspond to mostly relying on intracellular metal ion coordinating cysteine
the multicellular-specific metabolism and organization of the clusters.
extracellular matrix in general, and collagen fibrils in particular. At the organism level, the cellular functions involving redox-
These processes are clearly dominated by extracellular proteins, regulated conditionally disordered protein regions are typically
and are thus strongly associated with redox-sensitive disulfide embedded in a wide variety of developmental processes (Figure
bridge forming protein regions. 4D). Several of these processes are focused on either reproduc-
Analysis of network level processes shows that redox-sensitive tion, such as the formation of a labyrinthine layer, or the very
conditionally disordered proteins are heavily involved in the reg- early developmental stages of life, such as somitogenesis, gastru-
ulation of key processes (Figure 4B). Cell growth is governed lation, or endoderm formation. As these high level organismal
by several signals, and the processing of various growth factors processes typically rely on a large number of proteins, they gen-
(such as TGF-β) and retinoic acid is clearly dependent on con- erally do not show exclusivity toward either structural element.
ditional disorder, with a slight preference for disulfide bonded
proteins. In contrast, steroid hormone signal processing mainly
involves intracellular, metal ion coordinating structural switches.
Generic cell–cell communication involving various members of 3.6. Redox-Sensitive Structural Switches in Disease Related
the integrin receptor family relies on a balanced mix of disulfide Proteins
bonded and metal ion coordinating structural elements. A clear
exception to this rule is the interaction between nerve cells, as the As the identified human redox-sensitive proteins fulfill central
assembly and organization of synapses very heavily utilizes extra- biological roles, their modulation by mutations is expected to
cellular conditionally disordered proteins, stabilized by disulfide have serious physiological consequences. We used the Hum-
bonds. savar dataset of disease associated germline mutations collected
These molecular and network level processes enable charac- in Uniprot[54] to assess human pathogenic conditions associated
teristic cellular functions that are typically centered around cell with the perturbation of redox-sensitive conditionally disordered
movement, growth, differentiation, and proliferation (Figure regions. We collected disease mutations that either eliminate
4C). The majority of these processes require the coordination be- or introduce cysteine residues in the sequence. In general,
tween various pathways and utilize several molecular processes. cysteine is one of the least common amino acids in proteins.
As a result, these processes typically show very little preference Accordingly, the frequency of cysteine residue in sequences
for either of the two types of redox-sensitive structural switches. with diseases mutations is only 2.44%. However, its frequency
The two main exceptions to this rule are cell movement via among mutated residues is 6.85%, which places it as the second

Proteomics 2019, 1800070 1800070 (10 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

most mutation-prone residue (after arginine) in the disease mutations, disrupting highly conserved intrachain disulfide
dataset. We analyzed these cases to establish a link between the bonds, leading to the autoimmune disease periodic syndrome
induced structural changes, protein function, and phenotypic phenotypes. This perturbation could contribute to ligand-
alterations. independent and enhanced ligand-dependent tumor necrosis
The structural effect of cysteine modifying mutations shows a factor receptor signaling under oxidative stress,[65] although
clear correlation with the subcellular localization of the affected impaired cytokine receptor clearance was also proposed as a
protein. Extracellular regions are targeted through cysteines in- disease mechanism.[66]
volved in the formation of the stabilizing disulfide bond patterns, Disulfide bond pattern perturbation can also affect individual
while intracellular proteins are targeted through modulation of extracellular domains. The Ret receptor tyrosine kinase is mu-
metal ion binding (Table 2). Extracellular disulfide patterns can tated in the extracellular juxtamembrane cysteine-rich domain.
be disrupted by both the introduction and elimination of a cys- Similarly to NOTCH3, these mutations induce intermolecular
teine, as both lead to an uneven number of cysteines, resulting interactions between receptor molecules. The increased dimer-
in incorrect intra- or intermolecular bond formation. The native ization of Ret proteins leads to increased levels of autophospho-
disulfide patterns are typically highly conserved[55] and also con- rylation and tyrosine kinase activity,[67,68] underlying a frequent
tribute to protein stability and folding through indirect effects.[56] form of medullary thyroid cancer (Ret-MEN2A). Furthermore,
Mutations of such cysteines typically reduce the stability of the cysteine modulating mutations can also severely affect single
implicated domains, resulting in incorrect folding, or promoting domain extracellular proteins. A prime example is the arginine
aggregation, which affects several human plasma membrane re- vasopressin-neurophysin II (AVP-NPII), which is composed of a
ceptors and extracellular proteins. single domain, stabilized by a network of seven disulfide bonds.
One of the most targeted extracellular structural modules Both the creation or the abolishment of a Cys residue are likely to
in the analyzed dataset is EGF-like domains, which are stabi- alter the three-dimensional structure of the prohormone, which
lized via three pairs of cysteines forming three disulfide bonds. accumulates in the cell body, ultimately leading to neuronal
EGF-like domains are commonly found in tandem repeats in degeneration and hormonal deficit,[69] resulting in autosomal
the ligand binding regions of receptors, and are often targeted dominant familial neurohypophyseal diabetes insipidus.
by germline mutations. Cys-altering mutations in the EGF do- The disulfide-favoring oxidizing conditions are not only
mains of fibrillin, the major structural constituent of extracellu- present in the extracellular space, but can also be found in the
lar microfibrils,[57,58] results in connective tissue disorders such intermembrane space of the mitochondria. In the light of our
as Marfan syndrome. The perturbation results in short- or long- knowledge of redox-sensitive conditionally disordered regions lo-
range structural rearrangements and can affect the Ca2+ bind- cated in this compartment, COA6 provides an interesting case
ing ability of the domain. This can lead to increased susceptibil- of the disease mutations. The protein is involved in the matura-
ity to proteolysis, retention in the endoplasmic reticulum, and tion of the mitochondrial respiratory chain complex, and is criti-
delayed secretion.[57] It has been suggested that increased ox- cal for biogenesis of mtDNA-encoded COX2. A pathogenic muta-
idative stress can contribute to disease progression.[59] A radi- tion in COA6, leading to substitution of a conserved tryptophan
cally different phenotype is produced by EGF Cys mutations of for a cysteine residue, results in a loss of complex IV activity and
CRB1, linked to various retinal dystrophies. CRB1 has an impor- cardiomyopathy. Toward understanding the molecular basis of
tant role in maintaining retinal integrity, and its mutations re- pathogenesis, it was recently shown that the human COA6 W59C
sult in ROS accumulation and photoreceptor cell death.[60] Simi- mutant leads to an increased aggregation state or mislocalization
lar mutations in NOTCH3, cause the vascular dementia disease to the mitochondrial matrix.[70,71] In agreement with these obser-
Cerebral Autosomal Dominant Arteriopathy with Subcortical In- vations, the observed mutations introduce a redox-sensitive re-
farcts and Leukoencephalopathy (CADASIL). In this case, how- gion, which is not predicted in the native protein sequence.
ever, mutations lead to impaired intracellular trafficking and mat- In contrast to their extracellular counterparts, disease muta-
uration of the mutated Notch 3 receptors,[61] and are not directly tions affecting putative redox-sensitive conditionally disordered
related to oxidative stress.[62] In Notch3, the uneven number of regions located in proteins inside the cell target metal ion coor-
cysteines likely contributes to intermolecular disulfide bridges, dinating domains. The majority of such mutations perturb the
as these mutations increase multimerization and aggregate Zn2+ ion binding of various types of zinc finger (ZNF) domains,
formation.[63] and in these cases the mutations almost ubiquitously eliminate
Apart from EGF-like domains, disulfide pattern perturbation cysteines. As one of the most common functions of ZNF do-
by uneven number of cysteines was also linked to several other mains is DNA binding, several cysteine-modifying germline mu-
extracellular domains in tandem arrangements. The low-density tations affect transcription factors. One of the most well stud-
lipoprotein receptor (LDLR) contains a tandem array of LDL- ied such proteins is the human zinc-finger transcription factor
receptor domains and EGF-like domains. LDL-receptor domains WT1. WT1 ZNF cysteine mutations result in abnormal develop-
are responsible for ligand binding, and EGF-like domains are ment of the genitourinary system, and are associated with vari-
involved in the degradation and recycling of the receptor. Both ous diseases including Wilms tumors, Denys-Drash syndrome,
types of structural modules are common sites for Cys altering Nephrotic syndrome 4, and Meacham syndrome. The four C-
mutations, causing familial hypercholesterolemia.[64] Tumor terminal tandem ZNFs of WT1[72] are mutational hotspots, with
necrosis factor receptor superfamily member 1A (TNFRSF1A) the majority of the contained mutations altering the zinc coordi-
does not contain EGF-like domains, but its extracellular region nating cysteine residues of the ZNF2 and ZNF3 domains. These
is composed of four repeats of the TNFR domain, each stabilized mutations can increase the flexibility of the protein or alter its
by three disulfide bonds. TNRF can harbor Cys-abolishing DNA-binding specificity.[73] While there is no direct evidence that

Proteomics 2019, 1800070 1800070 (11 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

Table 2. Redox-sensitive regions affected by disease causing germline mutations. For each protein, the main disease-affected molecular processes,
affected signaling pathways/networks, and cellular/organismal functions are given based on associated GO terms, if available.

Gene name (UniProt Mutations in Associated disease Altered protein function Affected Pfam domain
accession) protein/redox-
sensitive regions

Extracellular and transmembrane proteins

FBN1 (P35555) 393/236 Marfan syndrome ECM organization; Response to EGF CA (disulfide bond)
various growth factor stimuli;
Regulation of osteoclast
development;
CRB1 (P82279) 82/21 Retinitis pigmentosa 12 Plasma membrane organization; Eye EGF, hEGF (disulfide bond)
photoreceptor cell development;
NOTCH3 (Q9UM47) 110/108 CADASIL1 Transcription initiation; EGF, EGF CA (disulfide bond)
Notch signaling pathway;
Neuron differentiation/fate
commitment;
LDLR (P01130) 104/34 Familial hypercholesterolemia Regulation of gene EGF CA,
expression/metabolic processes; FXa inhibition,Ldl recept a
Cellular response to lipid; (disulfide bond)
Cholesterol homeostasis;
TNFRSF1A (P19438) 13/8 Familial hibernian fever Regulation of transcription; TNFR c6 (disulfide bond)
I-kappaB kinase/NF-kappaB
signaling;
Inflammatory and immune response;
RET (P07949) 139/23 Multiple neoplasia type IIA Protein phosphorylation; N/A
Regulation of MAPK/integrin
pathways;
Neuron maturation/adhesion;
AVP-NPII (P01185) 39/18 Diabetes insipidus Regulation of gene expression; Hormone 5 (disulfide bond)
G-protein coupled receptor and
ERK1/2 signaling pathways;
Cell growth/proliferation;
Mitochondrial proteins
COA6 (Q5JTJ3) 2/1 Cardioencephalomyopathy, fatal Respiratory chain complex IV COX6B (disulfide bond)
infantile assembly;
Cytosolic and nuclear proteins
WT1 (P19544) 46/11 Denys-Drash syndrome, Meacham Regulation of zf-C2H2 (Zn2+ ion coordinating)
syndrome, Nephrotic syndrome 4 transcription/translation;
Regulation of apoptotic pathway;
Kidney/heart/gonad/adrenal gland
development;
PHF6 (Q8IWS0) 6/3 Boerjeson-Forssman-Lehmann Regulation of transcription; zf-HC5HC2H (Zn2+ ion
syndrome coordinating)
KMT2D (O14686) 13/3 Kabuki syndrome 1 Histone methylation/transcription zf-HC5HC2H 2 (Zn2+ ion
regulation; coordinating)
Response to estrogen;
Megakaryocyte differentiation/oocyte
growth;
PRKN (O60260) 30/5 Parkinson disease 2 Regulation of IBR (Zn2+ ion coordinating)
transcription/neurotransmitter
secretion/protein metabolism;
Response to unfolded protein,
ERAD/JNK/wnt pathway regulation;
Neural homeostasis/death;
LMX1B (O60663) 30/8 Nail-patella syndrome Regulation of transcription; LIM (Zn2+ ion coordinating)
Neuron differentiation;
FHL1 (Q13642) 13/3 Myopathy, X-linked, with postural Regulation of membrane LIM (Zn2+ ion coordinating)
muscle atrophy, Reducing body depolarization;
myopathy, X-linked 1B Muscle morphogenesis;

Proteomics 2019, 1800070 1800070 (12 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

WT1 is redox-regulated, in the case of the similar protein EGR1, it 4. Discussion

was shown that the redox state modules the DNA binding activity
of the protein.[74,75] Another transcription factor, PHF6 is involved Cysteine residues can serve as redox-regulated cellular switches
in chromatin regulation and neural development.[76] Germline modulating the biological activity of various proteins. It has been
mutations connected to the X-linked mental retardation disor- demonstrated in the case of a growing number of examples,
der Börjeson–Forssman–Lehmann syndrome eliminate the con- that such thiol switches may trigger conformational changes
served C45, C99, and C305 cysteines in either the C2HC or the in proteins, and promote a disorder-to-order (or order-to-
PHD-type zinc finger. These mutations affect the structure core disorder) transition upon changes in cellular redox conditions.
stability or zinc coordination of the PHF6-ePHD2 domain, and Unfortunately, the large-scale experimental discovery of such
destabilize the correct protein fold and consequently interfere thiol switches encoded in conditionally disordered proteins is
with the normal biological function the protein.[76] Apart from di- challenging and time consuming. Therefore, a bioinformatics
rect DNA binding, ZNF targeting mutations can affect transcrip- tool that points to the potential redox-regulated intrinsically
tion in a more indirect way. C1430R and C1471Y mutations in disordered regions can significantly speed up discovery of such
KMT2D likely disrupt the PHD finger fold, and reduce histone important regions at a proteome level, identifying interesting
binding and catalytic activity of the protein,[77] leading to Kabuki targets for further biochemical characterization. Here, we
syndrome. applied such a tool implemented in the IUPred2A web server[26]
The functional heterogeneity of ZNF domains results in the to assess the functional link between redox-regulated processes
germline modulation of redox-sensitive cysteines in proteins and the structural dynamics of the proteins involved.
with distinctively different biological functions as well. Muta- One of the main outcomes of the presented study is that redox-
tions in Parkin (PARK2), a ubiquitin ligase, are one of the pre- sensitive conditionally disordered protein regions (RSCDPs) are
dominant hereditary factors underlying some forms of Parkin- widespread in various proteomes, especially in eukaryotic mul-
son disease. Oxidative stress has been recognized as a ma- ticellular organisms. According to our predictions, around 5%
jor contributing factor to the disease.[78] At the protein level, of the proteome corresponds to proteins with redox-sensitive
a large fraction of the observed loss-of-function mutations in conditionally disordered protein regions in yeast, however, this
the hereditary form of the disease destabilize the protein by proportion increases to nearly 30% in the human proteome. This
affecting folding and stability of the complicated Zn2+ -bound highly unexpected result gains more credence with the domain-
architecture,[79,80] leading to impaired degradation of target pro- level analysis that established a strong correlation between the
teins, such as ZNF746 and BCL2.[81,82] Other proteins with redox- increased amount of RSCDPs and the expansion of potentially
sensitive cysteine mutations include FHL1 and LMX1B, both redox-sensitive domains, such as intracellular zinc fingers or
containing multiple LIM domains, which are typically involved extracellular EGF domains, at various points of evolution. In
in protein-protein interactions as opposed to DNA-binding. LIM yeast, the most common domains predicted to be redox-regulated
domains are generally 50–60 amino acids in size and share two through conditional disorder are various types of zinc fingers in-
characteristic zinc fingers, which are separated by two amino cluding the DNAJ C family, or certain subunits of the ribosome
acids.[83,84] LIM domain containing proteins have diverse biolog- (Table 1). In fruit fly species, the increase of predicted redox-
ical functions, often shuttling between the nucleus and the cy- regulated disordered regions can also be partially attributed to
toplasm, potentially even in a redox-regulated manner,[85] act- the expansion of the zinc finger families. However, novel do-
ing as adaptors or scaffolds to support the assembly of mul- mains also appear, such as the zinc binding LIM domains, some
timeric protein complexes, and can also contribute to the reg- of which were shown to change their intracellular location upon
ulation of localization.[83] Specifically, LMX1B controls the ex- redox regulation. The functional importance of these domains is
pression of key genes involved in mitochondrial functions.[86] further underlined by their involvement in various diseases that
The FHL1 gene is predominantly expressed in skeletal and car- alter cysteine residues with a critical role in this type of regula-
diac muscle, and its mutations are causative for several types tion. Moreover, another large group of predicted RSCDPs corre-
of hereditary myopathies. The Cys mutations are likely to sig- sponds to extracellular domains found only in multicellular or-
nificantly affect the structure and the folding of the proteins ganisms. The redox regulation of extracellular proteins could be
and have been shown to induce the formation of aggresome-like important for their transfer to the extracellular matrix, and as-
inclusions.[87] sembly in their target location. Thus, mutations of such criti-
The detailed examples are all modulated through either cal, redox-regulated cysteine residues can lead to devastating dis-
of the two main redox-sensitive structural elements: disul- eases (Table 2). Coupling of potentially redox-sensitive thiols and
fide bridges or metal ion clusters. However, the biologi- structural plasticity is very well characterized in proteins trans-
cal functions of these proteins vary, and as a consequence, ferred from the cytosol to other organelles, such as the mito-
their mutations lead to very heterogeneous phenotypic al- chondria and the peroxisome. Thus, a similar mechanism might
terations. These alterations are enabled by the distinct bio- be crucial for the biogenesis of extracellular proteins, includ-
logical processes and pathways in which these proteins are ing receptors. In addition to biogenesis, receptors could utilize
embedded. Table 2 also shows the relevant molecular, path- redox-regulated regions for sensing changes in redox status and
way/network, and cellular level processes of the mutated pro- mediate related signaling events.[65,67,68] Our results suggest
teins. In some cases the biological processes are not entirely that the human and fruit fly proteomes share many com-
clear, but the majority of the proteins do reflect the general mon redox-sensitive structural switching domains, correspond-
trends of processes utilizing redox-sensitive structural switches ing to similar functions. However, the abundance of these do-
(Figure 4). mains is further elevated in Homo sapiens, with the most drastic

Proteomics 2019, 1800070 1800070 (13 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

increase observed in the case of the C2H2 type zinc finger pathogens, including Trypanosoma congolense and Plasmodium
domains. falciparum. Our analysis suggests that these pathogens contain
This analysis supports a very well established interplay be- large fractions of proteins with redox-sensitive intrinsically dis-
tween protein- and redox homeostasis in cells,[8] revealing ordered regions (Figure 3A). Further analysis of these proteins
potential members of the protein homeostasis system that might might have therapeutic advances. In order to choose potential
employ redox-regulated conformational changes to regulate their candidates for experimental analysis, our method can be coupled
function during oxidative stress. One important group of RS- with the recently developed experimental tools, which enable
DCPs includes transcription factors, many of which contain real-time mapping of redox states in the parasite.[99]
zinc finger domains.[88] In yeast, it has been shown that mod- In conclusion, the results presented in this study show that
ulation of the redox status induces reversible changes in the many protein regions rely on metal binding cysteines or disul-
translational machinery and controls protein synthesis.[45] Exam- fide bonds for stability, which are potentially redox-regulated,
ples of redox-regulated proteins that rely on conditional disorder and involve structural transitions from an ordered state to a
are involved in mitochondrial transport and assembly. In mam- disordered state, or vice versa. The abundance of these types
malians, redox signaling via alteration of redox status of spe- of regions dramatically increases from yeast to human, and
cific protein thiols has been recognized as a major contributor involves a wide range of processes related to multicellularity,
to diverse key processes, such as stem cell proliferation,[89] verte- signaling, and regulation of transcription and translation. The
brate embryonic development,[90] neuronal development,[91] and presented analyses can direct further experimental study of
blood coagulation.[92] Redox regulation is also an emerging theme a crosstalk between redox regulation, structural changes, and
in cell–cell and cell–extracellular matrix interactions,[93] tying to- proteostasis related functions.
gether integrin signaling, angiogenesis, and various diseases. All
of these processes are present only in higher order eukaryotes,
and the heavy involvement of redox-sensitive conditionally disor-
Supporting Information
dered proteins is in line with the evolutionary trends identified
(Figure 3). Supporting Information is available from the Wiley Online Library or from
Altogether, these results reinforce the importance of redox the author.
regulation in key biological processes. However, the analysis of
human genetic variations also highlights that the mutations of
cysteine residues critical for such regulation are correlated with Acknowledgments
different pathologies (Table 2). This analysis shows that such
disease-related mutations are found in proteins with different This work was supported by the “Lendület” grant from the Hungar-
ian Academy of Sciences (LP2014-18) (Z.D., B.M., and G.E.), OTKA
functions and different cellular localization. It is tempting to
grant (K108798) (Z.D., B.M., and G.E.), the FIEK grant FIEK 16-1-6012-
speculate that some of these proteins might be involved in sens- 000 (Z.D.), the EMBO|EuropaBio fellowship 7544 (B.M.), the Israel Sci-
ing or maintaining redox homeostasis in the related organelles. ence Foundation (1765/13) (D.R.) and Human Frontier Science program
The understanding of the structural mechanisms underlying the (CDA00064/2014) (D.R.). The authors thank Meytal Radzinski for the crit-
development of these diseases has already served with potential ical reading and editing of the manuscript.
therapeutic options. CADASIL commonly arises via Notch3
cysteine mutations, and artificially introduced exon-skipping
removes the affected EGF-like domains, and can restore receptor Conflict of Interest
functionality in vitro.[94] The large-scale identification of similar
redox-sensitive protein regions, together with pathway/mutation The authors declare no conflict of interest.
analysis can serve as new targets for therapeutic intervention.
In addition to endogenous modulations leading to diseases,
RSCDPs and redox regulation play critical roles in host–pathogen Keywords
interactions as well. The role of redox-sensitive proteins in in-
teractions between viruses and eukaryotic hosts is known for conditional disorder, intrinsically disordered proteins, redox proteomics,
several examples, such as Hepatitis C virus,[95] coxsackievirus redox-sensitive structural switches, thiol switches
B3,[96] and Epstein–Barr virus.[97] Viruses, in general, exploit the
redox-state dependence of several signaling pathways of the host Received: September 9, 2018
Revised: December 13, 2018
cell, and this has even been recognized as having therapeutic Published online:
value.[98] Our results suggest that the use of redox-state depen-
dent proteins can be a generic theme among viruses targeting
multicellular hosts. Redox-state modulation is also an emerging
theme in the host-pathogen interactions of single-cell eukaryotic [1] R. van der Lee, M. Buljan, B. Lang, R. J. Weatheritt, G. W. Daugh-
drill, A. K. Dunker, M. Fuxreiter, J. Gough, J. Gsponer, D. T. Jones,
pathogens, such as trypanosoma, leishmania, plasmodium, and
P. M. Kim, R. W. Kriwacki, C. J. Oldfield, R. V. Pappu, P. Tompa,
others. Parasite survival and successful proliferation depend on V. N. Uversky, P. E. Wright, M. M. Babu, Chem. Rev. 2014, 114,
the capacity of the invading parasite to cope with the oxidation 6589.
blast of the host immune system. Therefore, an effective redox [2] N. S. Latysheva, T. Flock, R. J. Weatheritt, S. Chavali, M. M. Babu,
regulation mechanism is a crucial component for parasite Protein Sci. 2015, 24, 909.
survival that is found and characterized for many unicellular [3] P. E. Wright, H. J. Dyson, Nat. Rev. Mol. Cell Biol. 2015, 16, 18.

Proteomics 2019, 1800070 1800070 (14 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

[4] P. Tompa, Trends Biochem. Sci. 2002, 27, 527. [35] H.-S. Won, L. Y. Low, R. D. Guzman, M. Martinez-Yamout, U. Jakob,
[5] U. Jakob, R. Kriwacki, V. N. Uversky, Chem. Rev. 2014, 114, 6779. H. J. Dyson, J. Mol. Biol. 2004, 341, 893.
[6] D. Reichmann, U. Jakob, Curr. Opin. Struct. Biol. 2013, 23, 436. [36] P. C. F. Graf, M. Martinez-Yamout, S. VanHaerents, H. Lilie, H. J.
[7] B. Groitl, U. Jakob, Biochim. Biophys. Acta 2014, 1844, 1335. Dyson, U. Jakob, J. Biol. Chem. 2004, 279, 20529.
[8] D. Reichmann, W. Voth, U. Jakob, Mol. Cell 2018, 69, 203. [37] F. Arnesano, E. Balatri, L. Banci, I. Bertini, D. R. Winge, Structure 2005,
[9] A. J. Weids, S. Ibstedt, M. J. Tamás, C. M. Grant, Sci. Rep. 2016, 6, 13, 713.
24554. [38] B. Bloj, M. G. Galo, R. D. Morero, R. N. Farias, J. Nutr. 1976, 106,
[10] Q. Ding, E. Dimayuga, J. N. Keller, Antioxid. Redox Signal. 2006, 8, 1827.
163. [39] P. W. Kim, Z.-Y. J. Sun, S. C. Blacklow, G. Wagner, M. J. Eck, Science
[11] S. Reeg, T. Grune, Antioxid. Redox Signal. 2015, 23, 239. 2003, 301, 1725.
[12] P. D. Ray, B.-W. Huang, Y. Tsuji, Cell. Signal. 2012, 24, 981. [40] J. Zhou, S. Zhao, A. K. Dunker, J. Mol. Biol. 2018, 430, 2342.
[13] T. Finkel, J. Cell Biol. 2011, 194, 7. [41] J. M. Rice, A. Zweifach, M. A. Lynes, BMC Immunol. 2016, 17, 13.
[14] M. Radzinski, R. Fassler, O. Yogev, W. Breuer, N. Shai, J. Gutin, [42] S. W. Ragsdale, L. Yi, Antioxid. Redox Signal. 2011, 14, 1039.
S. Ilyas, Y. Geffen, S. Tsytkin-Kirschenzweig, Y. Nahmias, T. [43] P. M. Harrison, A. Kumar, N. Lang, M. Snyder, M. Gerstein, Nucleic
Ravid, N. Friedman, M. Schuldiner, D. Reichmann, ELife 2018, 7, Acids Res. 2002, 30, 1083.
https://doi.org/10.7554/eLife.37623. [44] N. Brandes, D. Reichmann, H. Tienson, L. I. Leichert, U. Jakob, J. Biol.
[15] J. Messens, J.-F. Collet, Antioxid. Redox Signal. 2013, 18, 1594. Chem. 2011, 286, 41893.
[16] H. Launay, P. Barré, C. Puppo, Y. Zhang, S. Maneville, B. Gontero, V. [45] U. Topf, I. Suppanz, L. Samluk, L. Wrobel, A. Böser, P. Sakowska, B.
Receveur-Bréchot, J. Mol. Biol. 2018, 430, 1218. Knapp, M. K. Pietrzyk, A. Chacinska, B. Warscheid, Nat. Commun.
[17] G. Ghag, C. J. Holler, G. Taylor, T. L. Kukar, V. N. Uversky, V. Ran- 2018, 9, 324.
gachari, Protein Sci. 2017, 26, 1759. [46] L. Popolo, E. Ragni, C. Carotti, O. Palomares, R. Aardema, J. W. Back,
[18] J. H. Hoffmann, K. Linke, P. C. F. Graf, H. Lilie, U. Jakob, EMBO J. H. L. Dekker, L. J. de Koning, L. de Jong, C. G. de Koster, J. Biol. Chem.
2004, 23, 160. 2008, 283, 18553.
[19] H. Fraga, J. Pujols, M. Gil-Garcia, A. Roque, G. Bernardo-Seisdedos, [47] H.-I. Choi, S. P. Lee, K. S. Kim, C. Y. Hwang, Y.-R. Lee, S.-K. Chae, Y.-S.
C. Santambrogio, J.-J. Bech-Serra, F. Canals, P. Bernadó, R. Grandori, Kim, H. Z. Chae, K.-S. Kwon, Free Radic. Biol. Med. 2006, 40, 651.
O. Millet, S. Ventura, Sci. Rep. 2017, 7, 16994. [48] F. Bonn, T. Tatsuta, C. Petrungaro, J. Riemer, T. Langer, EMBO J. 2011,
[20] B. McDonagh, S. Ogueta, G. Lasarte, C. A. Padilla, J. A. Bárcena, J. 30, 2545.
Proteomics 2009, 72, 677. [49] E. C. Curran, H. Wang, T. R. Hinds, N. Zheng, E. H. Wang, Sci. Rep.
[21] L. I. Leichert, F. Gehrke, H. V. Gudiseva, T. Blackwell, M. Ilbert, A. K. 2018, 8, 4630.
Walker, J. R. Strahler, P. C. Andrews, U. Jakob, Proc. Natl. Acad. Sci. [50] D. Prather, N. J. Krogan, A. Emili, J. F. Greenblatt, F. Winston, Mol.
USA 2008, 105, 8197. Cell. Biol. 2005, 25, 10122.
[22] M. Hillion, J. Bernhardt, T. Busche, M. Rossius, S. Maaß, D. Becher, [51] M. Rojas, G. W. Farr, C. F. Fernandez, L. Lauden, J. C. McCormack, S.
M. Rawat, M. Wirtz, R. Hell, C. Rückert, J. Kalinowski, H. Antelmann, L. Wolin, PLoS One 2012, 7, e52824.
Sci. Rep. 2017, 7, 1195. [52] O. Khalimonchuk, K. Rigby, M. Bestwick, F. Pierrel, P. A. Cobine, D.
[23] K. E. Menger, A. M. James, H. M. Cochemé, M. E. Harbour, E. T. R. Winge, Eukaryot. Cell 2008, 7, 1427.
Chouchani, S. Ding, I. M. Fearnley, L. Partridge, M. P. Murphy, Cell [53] J. M. Cherry, E. L. Hong, C. Amundsen, R. Balakrishnan, G. Binkley,
Rep. 2015, 11, 1856. E. T. Chan, K. R. Christie, M. C. Costanzo, S. S. Dwight, S. R. Engel,
[24] S. Rosenwasser, S. Graff van Creveld, D. Schatz, S. Malitsky, O. Tzfa- D. G. Fisk, J. E. Hirschman, B. C. Hitz, K. Karra, C. J. Krieger, S. R.
dia, A. Aharoni, Y. Levin, A. Gabashvili, E. Feldmesser, A. Vardi, Proc. Miyasato, R. S. Nash, J. Park, M. S. Skrzypek, M. Simison, S. Weng,
Natl. Acad. Sci. USA 2014, 111, 2740. E. D. Wong, Nucleic Acids Res. 2012, 40, D700.
[25] N. Brandes, H. Tienson, A. Lindemann, V. Vitvitsky, D. Reichmann, [54] M. L. Famiglietti, A. Estreicher, A. Gos, J. Bolleman, S. Géhant, L.
R. Banerjee, U. Jakob, ELife 2013, 2, e00306. Breuza, A. Bridge, S. Poux, N. Redaschi, L. Bougueleret, I. Xenarios,
[26] B. Mészáros, G. Erdos, Z. Dosztányi, Nucleic Acids Res. 2018, 46, UniProt Consortium, Hum. Mutat. 2014, 35, 927.
W329. [55] R. R. Thangudu, M. Manoharan, N. Srinivasan, F. Cadet, R. Sowd-
[27] Z. Dosztányi, V. Csizmók, P. Tompa, I. Simon, J. Mol. Biol. 2005, 347, hamini, B. Offmann, BMC Struct. Biol. 2008, 8, 55.
827. [56] R. Bhattacharyya, D. Pal, P. Chakrabarti, Protein Eng. Des. Sel. 2004,
[28] G. Wang, R. L. Dunbrack Jr, Nucleic Acids Res. 2005, 33, W94. 17, 795.
[29] T. UniProt Consortium, Nucleic Acids Res. 2018, 46, 2699. [57] K. A. Zeyer, D. P. Reinhardt, Mutat. Res. Rev. Mut. Res. 2015, 765, 7.
[30] R. D. Finn, P. Coggill, R. Y. Eberhardt, S. R. Eddy, J. Mistry, A. L. [58] P. A. Handford, Biochim. Biophys. Acta 2000, 1498, 84.
Mitchell, S. C. Potter, M. Punta, M. Qureshi, A. Sangrador-Vegas, [59] F. Romaniello, D. Mazzaglia, A. Pellegrino, S. Grego, R. Fiorito, A.
G. A. Salazar, J. Tate, A. Bateman, Nucleic Acids Res. 2016, 44, Ferlosio, L. Chiariello, A. Orlandi, Cardiovasc. Pathol. 2014, 23, 261.
D279. [60] F. J.-M. Chartier, É. J.-L. Hardy, P. Laprise, J. Cell Biol. 2012, 198, 991.
[31] R. P. Huntley, T. Sawford, P. Mutowo-Meullenet, A. Shypitsyna, C. [61] H. Karlström, P. Beatus, K. Dannaeus, G. Chapman, U. Lendahl, J.
Bonilla, M. J. Martin, C. O’Donovan, Nucleic Acids Res. 2015, 43, Lundkvist, Proc. Natl. Acad. Sci. USA 2002, 99, 17119.
D1057. [62] J. Campolo, R. De Maria, C. Mariotti, C. Tomasello, M. Parolini, M.
[32] I. Janda, Y. Devedjiev, U. Derewenda, Z. Dauter, J. Bielnicki, D. R. Frontali, D. Inzitari, R. Valenti, A. Federico, F. Taroni, O. Parodi, PLoS
Cooper, P. C. F. Graf, A. Joachimiak, U. Jakob, Z. S. Derewenda, Struc- One 2013, 8, e67077.
ture 2004, 12, 1901. [63] M. Duering, A. Karpinska, S. Rosner, F. Hopfner, M. Zechmeister, N.
[33] B. Groitl, S. Horowitz, K. A. T. Makepeace, E. V. Petrotchenko, C. H. Peters, E. Kremmer, C. Haffner, A. Giese, M. Dichgans, C. Opherk,
Borchers, D. Reichmann, J. C. A. Bardwell, U. Jakob, Nat. Commun. Hum. Mol. Genet. 2011, 20, 3256.
2016, 7, 10357. [64] D.-W. Zhang, T. A. Lagace, R. Garuti, Z. Zhao, M. McDonald, J.
[34] D. Reichmann, Y. Xu, C. M. Cremers, M. Ilbert, R. Mittelman, M. C. D. Horton, J. C. Cohen, H. H. Hobbs, J. Biol. Chem. 2007, 282,
Fitzgerald, U. Jakob, Cell 2012, 148, 947. 18602.

Proteomics 2019, 1800070 1800070 (15 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.proteomics-journal.com

[65] H. Z. Ozsoy, N. Sivasubramanian, E. D. Wieder, S. Pedersen, D. L. [82] D. Chen, F. Gao, B. Li, H. Wang, Y. Xu, C. Zhu, G. Wang, J. Biol. Chem.
Mann, J. Biol. Chem. 2008, 283, 23419. 2010, 285, 38214.
[66] M. F. McDermott, I. Aksentijevich, J. Galon, E. M. McDermott, B. W. [83] Q. Zheng, Y. Zhao, Biol. Cell 2007, 99, 489.
Ogunkolade, M. Centola, E. Mansfield, M. Gadina, L. Karenko, T. Pet- [84] J. L. Kadrmas, M. C. Beckerle, Nat. Rev. Mol. Cell Biol. 2004, 5,
tersson, J. McCarthy, D. M. Frucht, M. Aringer, Y. Torosyan, A. M. 920.
Teppo, M. Wilson, H. M. Karaarslan, Y. Wan, I. Todd, G. Wood, R. [85] M. Shibanuma, K. Mori, J.-R. Kim-Kaneyama, K. Nose, Antioxid. Redox
Schlimgen, T. R. Kumarajeewa, S. M. Cooper, J. P. Vella, C. I. Amos, Signal. 2005, 7, 335.
J. Mulley, K. A. Quane, M. G. Molloy, A. Ranki, R. J. Powell, G. A. Hit- [86] H. Doucet-Beaupré, C. Gilbert, M. S. Profes, A. Chabrat, C. Pacelli, N.
man, J. J. O’Shea, D. L. Kastner, Cell 1999, 97, 133. Giguère, V. Rioux, J. Charest, Q. Deng, A. Laguna, J. Ericson, T. Perl-
[67] A. Corcoran, T. G. Cotter, FEBS J. 2013, 280, 1944. mann, S.-L. Ang, F. Cicchetti, M. Parent, L.-E. Trudeau, M. Lévesque,
[68] M. Kato, T. Iwashita, K. Takeda, A. A. Akhand, W. Liu, M. Yoshihara, Proc. Natl. Acad. Sci. USA 2016, 113, E4387.
N. Asai, H. Suzuki, M. Takahashi, I. Nakashima, Mol. Biol. Cell 2000, [87] J. Schessl, Y. Zou, M. J. McGrath, B. S. Cowling, B. Maiti, S. S. Chin,
11, 93. C. Sewry, R. Battini, Y. Hu, D. L. Cottle, M. Rosenblatt, L. Spruce, A.
[69] S. Baglioni, G. Corona, M. Maggi, M. Serio, A. Peri, Eur. J. Endocrinol. Ganguly, J. Kirschner, A. R. Judkins, J. A. Golden, H.-H. Goebel, F.
2004, 151, 605. Muntoni, K. M. Flanigan, C. A. Mitchell, C. G. Bönnemann, J. Clin.
[70] D. A. Stroud, M. J. Maher, C. Lindau, F.-N. Vögtle, A. E. Frazier, E. Invest. 2008, 118, 904.
Surgenor, H. Mountford, A. P. Singh, M. Bonas, S. Oeljeklaus, B. [88] K. S. Eom, J. S. Cheong, S. J. Lee, J. Microbiol. Biotechnol. 2016, 26,
Warscheid, C. Meisinger, D. R. Thorburn, M. T. Ryan, Hum. Mol. 2019.
Genet. 2015, 24, 5404. [89] C. L. Bigarella, R. Liang, S. Ghaffari, Development 2014, 141, 4206.
[71] D. Pacheu-Grau, B. Bareth, J. Dudek, L. Juris, F.-N. Vögtle, M. Wissel, [90] A. R. Timme-Laragy, M. E. Hahn, J. M. Hansen, A. Rastogi, M. A. Roy,
S. C. Leary, S. Dennerlein, P. Rehling, M. Deckers, Cell Metab. 2015, Semin. Cell Dev. Biol. 2018, 80, 17.
21, 823. [91] M. Olguı́n-Albuerne, J. Morán, Antioxid. Redox Signal. 2018, 28, 1603.
[72] D. Wang, J. R. Horton, Y. Zheng, R. M. Blumenthal, X. Zhang, X. [92] O. Herkert, T. Djordjevic, R. S. BelAiba, A. Görlach, Antioxid. Redox
Cheng, Nucleic Acids Res. 2018, 46, 3864. Signal. 2004, 6, 765.
[73] F. Borel, K. C. Barilla, T. B. Hamilton, M. Iskandar, P. J. Romaniuk, [93] F. Buricchi, E. Giannoni, G. Grimaldi, M. Parri, G. Raugei, G. Ram-
Biochemistry 1996, 35, 12070. poni, P. Chiarugi, Cell Adh. Migr. 2007, 1, 33.
[74] A. Pines, N. Bivi, M. Romanello, G. Damante, M. R. Kelley, E. D. [94] J. W. Rutten, H. G. Dauwerse, D. J. M. Peters, A. Gold-
Adamson, P. D’Andrea, F. Quadrifoglio, L. Moro, G. Tell, Free Radic. farb, H. Venselaar, C. Haffner, G.-J. B. van Ommen, A. M.
Res. 2005, 39, 269. Aartsma-Rus, S. A. J. Lesnik Oberstein, Brain 2016, 139,
[75] R. Ammendola, M. Mesuraca, T. Russo, F. Cimino, Eur. J. Biochem. 1123.
1994, 225, 483. [95] V. C. Chu, S. Bhattacharya, A. Nomoto, J. Lin, S. K. Zaidi, T. D. Oberley,
[76] Z. Liu, F. Li, K. Ruan, J. Zhang, Y. Mei, J. Wu, Y. Shi, J. Biol. Chem. 2014, S. A. Weinman, S. Azhar, T.-T. Huang, PLoS One 2011, 6, e28551.
289, 10069. [96] M. A. Beck, J. Handy, O. A. Levander, Ann. N. Y. Acad. Sci. 2000, 917,
[77] M. Ali, R. A. Hom, W. Blakeslee, L. Ikenouye, T. G. Kutateladze, 906.
Biochim. Biophys. Acta 2014, 1843, 366. [97] P. Wang, L. Day, J. Dheekollu, P. M. Lieberman, J. Virol. 2005, 79,
[78] O. Hwang, Exp. Neurobiol. 2013, 22, 11. 13298.
[79] T. Wauer, D. Komander, EMBO J. 2013, 32, 2099. [98] L. Nencioni, R. Sgarbanti, D. Amatore, P. Checconi, I. Celestino, D.
[80] S. A. Beasley, V. A. Hristova, G. S. Shaw, Proc. Natl. Acad. Sci. USA Limongi, S. Anticoli, A. T. Palamara, E. Garaci, Curr. Pharm. Des. 2011,
2007, 104, 3095. 17, 3898.
[81] J.-H. Shin, H. S. Ko, H. Kang, Y. Lee, Y.-I. Lee, O. Pletinkova, J. C. [99] M. Rahbari, S. Rahlfs, E. Jortzik, I. Bogeski, K. Becker, PLoS One 2017,
Troconso, V. L. Dawson, T. M. Dawson, Cell 2011, 144, 689. 12, e0174837.

Proteomics 2019, 1800070 1800070 (16 of 16) 

C 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim