MINIREVIEW
Flexible nets
The roles of intrinsic disorder in protein interaction networks
A. Keith Dunker
1
, Marc S. Cortese
1
, Pedro Romero
1,2
, Lilia M. Iakoucheva
3
and
Vladimir N. Uversky
1,4
1 Department of Biochemistry and Molecular Biology, and the Center for Computational Biology and Bioinformatics, Indiana University
School of Medicine, Indianapolis, IN, USA
2 School of Informatics, Indiana University – Purdue University Indianapolis, IN, USA
3 Laboratory of Statistical Genetics, The Rockefeller University, New York, NY, USA
4 Institute for Biological Instrumentation, Russian Academy of Sciences, Moscow Region, Russia
Scale-free networks and hubs
In biological systems, processes such as growth, energy
generation, cell division and signaling are integrated by
large, intricate networks. These biological networks, as
well as certain nonbiological networks, especially those
involved in communications such as the internet and
cellular phone systems, are classified as scale-free net-
works (SFNs) [1–3]. The basic feature that separates
these networks from non-SFNs such as regular
Correspondence
A.K. Dunker, Department of Biochemistry
and Molecular Biology, and the Center for

CaM, calmodulin; Cdk, cyclin-dependent protein kinase; CKI, Cdk inhibitor protein; GSK3b, glycogen synthase kinase 3 beta; ID, intrinsically
disordered; MoRE, molecular recognition element; NER, nucleotide excision repair; PDB, Protein Data Bank; PONDRÒ, predictors of naturally
disordered regions; RGN, regular network; RNN, random network; SFN, scale-free network; XPA, xeroderma pigmentosum group A protein;
FRAT, frequently rearranged in advanced T-cell lymphomas; Wnt, wingless type MMTV integration site family; HMG, high mobility group;
VL-XT, a predictor of intrinsic disorder that integrates various methods-based predictor of long disordered regions and X-ray based N-andC-
terminal predictors; VSL1, length-dependent predictor of intrinsic protein disorder; RPA, replication protein A; ERCC1, exchange repair cross
complementing complex 1; TFIIH, transcription factor IIH; XAB, XPA binding protein; p27
Kip
, cyclin-dependent kinase inhibitor protein p27/1B.
FEBS Journal 272 (2005) 5129–5148 ª 2005 FEBS 5129
networks (RGNs) or random networks (RNNs) is the
presence of hubs. Hubs are highly connected nodes
that have hundreds, thousands or even millions of
links [1,4]. The existence of hubs and their frequency
impart two features to SFNs that provide substantial
benefit to large complex networks: (a) increased
robustness with regard to random defects and (b) shor-
ter distances (in terms of the number of intervening
nodes) between any two points [5]. RGNs are grid-like
with invariant node connectivity, whereas RNNs are
characterized by stochastic variations in node connec-
tivity [5,6]. Despite the random placement of links in
RNNs, the vast majority of nodes still have approxi-
mately the same connectivity.
Figure 1A,B compare an RNN to a similar-sized
SFN to illustrate an important property of the latter.
In SNFs, the hub nodes are connected to a dramatic-
ally greater fraction of all nodes than the nodes with
high connectivity in RNNs [7]. This provides the abil-
ity for a signal to travel from any node to any other

same number of nodes (130) and links
(430). (Reproduced from [205] with the per-
mission of the author, ª Institute of Physics
and IOP Publishing Ltd., 2000–05.) (C) Yeast
protein interaction network map. The color
of a node signifies the phenotypic effect of
removing the corresponding protein (red,
lethal; green, nonlethal; orange, slow
growth; yellow, unknown). (Modified from
[9] with the permission of the authors,
ª Nature Publishing Group, 1998–2005).
Flexible nets A. K. Dunker et al.
5130 FEBS Journal 272 (2005) 5129–5148 ª 2005 FEBS
derived protein–protein interaction network (Fig. 1C)
suggests a similar underlying architecture.
The scale-free nature of protein–protein interaction
networks gives them the advantages of high connectiv-
ity and robustness. For example, the connectedness of
RNNs decays steadily as nodes fail in a random fash-
ion. The surviving network breaks into progressively
smaller and increasingly separate subnets that lose the
ability to communicate with one another. On the other
hand, SFNs show much less degradation from random
node failure because highly connected hubs serve to
maintain the integrity of the network. Because SFN
hubs comprise a small fraction of total nodes, they are
statistically less likely to fail as a result of random
deleterious events. Although the error tolerance of
SFNs is high, it is important to note that the failure of
hubs quickly leads to the breakdown of connectivity

throughout full interactomes remains to be verified.
Secondly, experimental protein–protein interaction
data contains a significant fraction of false negatives
and positives. This has been illustrated by studies com-
paring the results from various large-scale efforts [15–
18]. Increasing the quality of existing data can be
addressed by comparing and combining datasets and
adding additional methods of analysis such as gene
neighborhood, gene co-occurrence, gene fusion events,
mRNA expression correlations and lethality of knock-
outs [19]. Despite the above described limitations, ana-
lysis of existing protein–protein interaction data can
lead to useful information.
Comparison of protein–protein interaction networks
across species has significant potential for the study of
molecular evolution. Useful tools have been construc-
ted for the comparison of protein–protein interaction
networks across species [20–23], and interesting evolu-
tionary inferences are being made, such as the observa-
tion that proteins having a more central position in
the networks of different species (i.e., hubs) appear to
evolve more slowly and are evidently more likely to be
essential for survival. These observations are consistent
with Fisher’s classic proposal that pleiotropy con-
strains evolution [24,25]. Other important considera-
tions include the timing and the locations of the hub
protein interactions. Some hub proteins have multiple
simultaneous interactions (party hubs), while others
have multiple sequential interactions separated in time
or in space (date hubs) [26]. It has been suggested that

FEBS Journal 272 (2005) 5129–5148 ª 2005 FEBS 5131
Many proteins have been shown to exist under
apparently physiological conditions as dynamic ensem-
bles. Instead of having relatively fixed bonds and
angles as in structured proteins, the backbone bonds
and angles of such proteins vary significantly over
time, with no specific equilibrium values while under-
going noncooperative conformational changes. In
other words, such proteins or protein regions do not
have rigid 3D structure under physiological conditions
in vitro [31–47]. Furthermore, these intrinsically disor-
dered (ID) proteins and regions are known to carry
out numerous biological functions including cell signa-
ling [35], molecular recognition [48], and various
other interactions with proteins and nucleic acids
[32,34,35,37,39–43,49–51].
Recently, a number of groups have published predic-
tors of protein disorder, several of which are available
on the web (reviewed in [48]; see also http://www.
disprot.org). These predictors are based on the
assumption that the absence of rigid structure is enco-
ded in specific features of the amino acid sequence
[52,53]. In fact, statistical analysis shows that amino
acid sequences encoding for ID proteins or regions are
significantly different from those of ordered proteins
on the basis of local amino acid composition, flexibil-
ity, hydropathy, charge, coordination number and
several other factors [34,52,54–56]. A signature of a
probable ID region is the presence of low sequence
complexity coupled with amino acid compositional

Qualitatively, it seems reasonable that unstructured
proteins could serve as hubs, providing a simpler basis
for responding to changes in the environment as
compared to rigid proteins. For example, disordered
regions can bind partners with both high specificity
and low affinity [65], suggesting that disorder-based
signaling and regulatory interactions can be highly spe-
cific but be easily reversed. These capabilities meet the
fundamental requirements of signaling interactions –
specificity and reversibility [49] with minimal structural
requirements. Another crucial property of ID proteins
and regions that could contribute to their function in
signaling networks is binding diversity; i.e., their ability
to partner with many other proteins and other ligands,
such as nucleic acids [66]. This opens the possibility of
disordered regulatory regions that are capable of bind-
ing many different partners.
An interesting consequence of the capability of ID
proteins and regions to interact with different binding
partners is the potential for polymorphism in the
bound state. That is, such proteins could have com-
pletely different geometries in the rigidified structures
that are induced upon binding to different partners
[48]. This conjecture has been confirmed at atomic
resolution. Portions of axin and frequently rearranged
in advanced T-cell lymphoma protein (FRAT), which
possess negligible sequence similarity, both interact
with an intrinsically disordered loop of glycogen syn-
thase kinase 3 beta (GSK3b) that adopts ordered
structure upon binding [67]. The binding sites for the

tions. By this means, ID hub proteins and regions
could serve multiple and distinct signaling networks
and be regulated via different pathways. For example,
GSK3b plays a crucial role in the wingless-type
MMTV integration site family (Wnt) signaling path-
way by controlling the levels of b-catenin [69–71], and
GSK3b is also known to be involved in insulin and
growth factor signaling pathways [72–75]. GSK3b
functions as a signal transducer for these two com-
pletely independent pathways without any obvious
cross-talk or interference [67]. In the Wnt signaling
network, a subset of the cellular GSK3b pool is incor-
porated into a multiprotein complex that brings
GSK3b and its b-catenin substrate into close proxim-
ity. In the insulin signaling pathway, GSK3b operates
via a completely different mechanism, where the phos-
phorylation of Ser9 converts the disordered N-termi-
nus of GSK3b to an autoinhibitory segment that
blocks access to the active site substrate binding cleft
[76]. The functional segregation of the insulin and Wnt
signaling networks requires either the absence of an
exchange between the subsets of the cellular GSK3b
molecules involved in each pathway, or suggests
mutual exclusion of the two processes. That is, the
involvement of GSK3b with the axin–adenomatous
polyposis coli complex can reverse (via the action of
the specific phosphatases associated with the men-
tioned complex [77,78]) or override the inhibitory Ser9
phosphorylation present on a recruited GSK3b mole-
cule via the substantial enhancement in activity

protein A (HMGA), and synaptobrevin (Table 1). An
interesting, well-studied, illustrative example of this
group of hubs is provided by HMGA [formerly called
HMG-I(Y)], a founding member of a new protein class
called architectural transcription factors [79]. As dis-
cussed in more detail below, this protein has been
implicated in the development of cancer and several
other pathological conditions [80]. HMGA is consid-
ered a central hub of nuclear function, being able to
bind to at least 18 known protein partners as well as
to several specific DNA structures [80].
Both circular dichroism (CD) [81] and nuclear mag-
netic resonance (NMR) [82,83] indicate that HMGA
lacks structure, with the molecule exhibiting a random
coil-like structure over its entire length. The atypical
electrophoretic mobility of this molecule [84] also sug-
gests a high content of extended structure. Figure 2A
compares the results of PONDRÒ analysis by two pre-
dictors of intrinsic disorder, firstly, a predictor of intrin-
sic disorder that integrates various methods-based
predictor of long disordered regions and X-ray based N-
and C-terminal predictors (VL-XT) [52,57,59] and sec-
ondly, a length-dependent predictor of intrinsic protein
disorder (VSL1) [85]. While VL-XT is the most well-
characterized member of the PONDRÒ family, VSL1
is more accurate overall and, indeed, obtained the best
results of the 20 order ⁄ disorder predictors tested in the
6th Critical Assessment of Methods for Protein Struc-
ture Prediction (CASP6; />In complete agreement with the experimental data, the
predictor outputs in Fig. 2A indicate that the HMGA

BRCA1 (P38398) 170–1649 [169] 79 158 ⁄ 14 16 ⁄ 011976⁄ 8 p53, ATM, BRCA2, c-Myc [169];
Chk1 & 2 [170]
XPA (P23025) 1–102, 210–273 [113] 63 27 ⁄ 24⁄ 04112⁄ 0 RPA70, RPA34, ERCC1, TFIIH,
XAB1 & 2 [171]
Estrogen receptor
a (P03372)
1–184 [172] 31 69 ⁄ 612⁄ 011690⁄ 1 p53 [173]; BRCA1 [174]; TATA box
binding protein [172]; calmodulin,
c-Jun [175]
p53 (P04637) 1–73 [176]; 183–188, 224–227 [177];
291–312 [178]; 319–323, 357–360 [179]
29 1900 ⁄ 40 34 ⁄ 0 239 164 ⁄ 8 Mdm2, ATM, ERK, p38, BCL-X
L
[180]
Mdm2 (Q00987) 1–17 [181]; 17–24, 110– 125 [182];
210–304 [183]
26 95 ⁄ 411⁄ 07229⁄ 3 p53, ARF, ATM, CK2, HIF-1a [184]
Calcineurin, subunit A
(Q08209)
1–13, 390–414, 469–521 [185] 16 451 ⁄ 24 1 ⁄ 0315⁄ 1 NFAT [186]; calcipressin 1 [187];
cabin 1 [188]; SOCS-3 [189];
calsarcin [190]
14-3-3¢n (P63104) 68–77, 134–137, 230–245 [191]
a
12 29 ⁄ 61⁄ 09761⁄ 1 p53, Wee1, Tau, Raf-1 Cdc25C,
Bad [192]
Cdk2 (P24941) 36–46, 152–162 [121]
a
7322⁄ 611⁄ 012530⁄ 15 protein phosphatase 2 A, cyclin E1,
DNA polymerase alpha [193]; BRCA1,

least 18 different transcription factors have been repor-
ted to specifically interact with HMGA proteins (sum-
marized in [80]). The list of proteins known to interact
with HMGA includes transcription factors such as
AP-1, ATF-2 ⁄ c-Jun heterodimer, IRF-1, c-Jun, NF-jB
p50 ⁄ p65 heterodimer, C ⁄ EBPb, Elf-1, NF-AT, NF-jB
A
B
C
Fig. 2. Order ⁄ disorder predictions on three
hub proteins. (A) PONDRÒ VL-XT (red) and
VSL1 (magenta) predictions on the human
HMGA protein sequence (Swiss-Prot acces-
sion number P17096). Green horizontal bars
correspond to the areas of the protein that
have been identified as the minimal regions
required for specific protein–protein interac-
tions with other transcription factors (after
[80]): 1, IRF-1; 2, ATF-1 ⁄ c-Jun; 3, NF-Y; 4,
SRF; 5, NF-jB; 6, p50; 7, Tst-1 ⁄ Oct-6; 8,
HIPK-2. Although only eight target proteins
are shown here, it has been established that
HMGA physically interacts with at least 18
transcription factors [80]. Dark yellow hori-
zontal bars correspond to the areas of the
protein (known as AT-hooks) that are
involved in DNA binding. A PONDRÒ
score ¼ 0.5 corresponds to a prediction of
disorder. (B) PONDRÒ VL-XT (red) and VSL1
(magenta) predictions on the Xenopus laevis

p50 homodimer, NF-jB p65, NF-Y, Oct-1, Oct-2 A,
PIAS3, Pu.1, RNF4, SRF, and Tst-1 ⁄ Oct-6 hetero-
dimer [80].
HMGA gene expression is maximal during embry-
onic development [93] and has been suggested to be
involved in the control of cell growth and differenti-
ation [94]. Interestingly, overproduction of HMGA
can be oncogenic and promote tumor progression and
metastasis via dramatic alterations in numerous signa-
ling pathways [80]. Based on these observations, it was
suggested that HMGA proteins function in the cell as
highly connected ‘nodes’ of protein–DNA and pro-
tein–protein interactions that influence a diverse array
of normal biological processes including growth, pro-
liferation, differentiation and death [80].
In summary, HMGA is a well-studied hub protein
that in the absence of binding partners has been
characterized to be completely disordered by NMR
[82,83] and CD [81], supporting the hypothesis that
hub proteins are strong candidates to possess appreci-
able amounts of disorder. The HMGA example was
also a major factor leading to the suggestion that
hub proteins as a group might depend on intrinsic
disorder [49]. This supposition is supported below by
several additional examples of hub proteins that util-
ize ID regions in their associations with multiple
partners.
Partially disordered hub proteins
Table 1 also lists several hub proteins that possess an
intermediate range of both ordered and disordered seg-

that is responsible for localizing XPA in the nucleus
[105]; the acidic region (residues 78–84) that is import-
ant for interaction with ERCC1 [106,107]; and the
C-terminal region (residues 226–273) that binds to
TFIIH [108,109]. Furthermore, the central region (resi-
dues 98–219) is the minimal polypeptide that preferen-
tially binds damaged DNA [110]. Finally, the fragment
98–187 is necessary for binding to the 70 kDa subunit
of RPA [103,104]. Figure 2B presents the distribution
of the PONDRÒ VL-XT and VSL1 scores within the
XPA sequence and illustrates the long predicted disor-
dered regions at or near the two ends (from M1 to
A55 and from S63 to P88 at the N-terminus and from
L183 to E230 near the C-terminus). Importantly,
Fig. 2B shows that the central DNA-binding domain
is likely to be mostly ordered, whereas the multiple
protein binding sites are located in the regions that are
likely to be disordered.
In agreement with the predictions (Fig. 2B), NMR
solution structure of a human XPA fragment contain-
ing the minimal DNA-binding domain (residues
98–219), revealed that one-third of this molecule is dis-
ordered [111,112]. These conclusions were further con-
firmed by the results of limited proteolysis analysis
[113]: mild trypsin digestion produced cuts at 33 of the
possible 48 sites, with no cleavage at any of the 14
possible sites in the internal DNA-binding region
(Q85–I179) (Fig. 2B). The observed cleavage sites were
predominantly in two of the large regions of predicted
disorder [113]. In general, it is believed that cut sites

forming active heterodimeric complexes. Eight Cdk
family members (Cdk1–Cdk8) and nine cyclins (A–I)
have been identified so far. Interestingly, each Cdk pairs
with a separate cyclin class, most of which have at least
two members [117,118]. For example, Cdk1 together
with cyclin B1 directs the G2 ⁄ M transition. Exit from
G1, in contrast, is primarily under the control of cyclin
D ⁄ Cdk4 ⁄ 6. Finally, two other cyclins (A and E) that
pair with Cdk2 are required for the G1 ⁄ S transition and
progression through the S phase [117,118].
All Cdks have similar sizes (30–40 kDa) and share at
least 40% sequence identity, including the highly con-
served 300 residue catalytic core. On the contrary, the
cyclin subunits vary in size (30–80 kDa), but all contain
a homologous 100 residue cyclin box domain. The Cdk
subunits are not catalytically active unless they bind to
a cyclin partner and form a basally active complex.
The fully active complex is produced when the Cdk is
phosphorylated [116,119]. Crystal structures of several
Cdks (phosphorylated and dephosphorylated) and their
complexes with cyclins and inhibitors have been solved
[119]. All Cdks have the same overall fold as other
eukaryotic protein kinases. For example, monomeric
Cdk2 consists of an N-terminal lobe rich in b-sheet (N
lobe), a larger C-terminal lobe rich in a-helix (C lobe),
and a deep cleft at the junction of the two lobes that is
the site of ATP binding and catalysis [120]. Figure 2C
represents distribution of the PONDRÒ VL-XT and
VSL1 scores within the human Cdk2 sequence and
illustrates that this protein is likely to be almost com-

Kip1
inactivate Cdk2 and Cdk4 cyclin complexes by
binding to them, whereas p16
INK4
and p15
INK4B
are
specific for Cdk4 and Cdk6. Importantly, CKIs are able
to inhibit Cdk–cyclin activity by blocking formation of
active Cdk–cyclin complexes via binding to inactive
Cdk or by binding to the active complex [116,119].
The Cdk inhibitor p21
Waf1 ⁄ Cip1 ⁄ Sdi1
is important for
p53-dependent cell cycle control, mediating G1 ⁄ S
arrest through inhibition of Cdks and possibly through
inhibition of DNA replication [123]. A striking dis-
order-to-order transition for p21
Waf1 ⁄ Cip1 ⁄ Sdi1
upon
binding to one of its biological targets, Cdk2, was
demonstrated by proteolytic mapping, CD spectropo-
larimetry, and NMR spectroscopy [66]. In fact, it has
been established that p21
Waf1 ⁄ Cip1 ⁄ Sdi1
and its NH
2
-ter-
minal fragment, being active as Cdk inhibitors, lacked
stable secondary or tertiary structure in free solution.

Cdk-
inhibition domain (residues 22–106) bound to human
cyclin A–Cdk2 complex shows that residues 25–93 of
A. K. Dunker et al. Flexible nets
FEBS Journal 272 (2005) 5129–5148 ª 2005 FEBS 5137
p27
Kip1
bind in an ordered conformation comprising
an a-helix, a 3
10
helix, and b-structure [120]. Import-
antly, the p27
Kip1
Cdk-inhibition domain was shown
to lack an intramolecular hydrophobic core. Instead,
p27
Kip1
interacts with the cyclin A–Cdk2 complex as
an extended structure, being bound to both cyclin A
and Cdk2. On cyclin A, it interacts with a groove
formed by conserved cyclin box residues. On Cdk2,
p27
Kip1
binds and rearranges the amino-terminal lobe
and also inserts into the catalytic cleft, mimicking ATP
in the context of the cyclin A–Cdk2 complex [120]. In
contrast, the unbound p27
Kip1
Cdk-inhibition domain
is intrinsically disordered (natively unfolded) as shown

and 2) are unstructured and flexible before binding
Cdk–cyclin complexes. The staple analogy is completed
by a linker helix (partially structured in the unbound
state) that connects the two prongs. This analogy is
illustrated in Fig. 3A, which presents a model for
p27
Kip1
binding to the Cdk2–cyclin A binary complex
and shows the importance of both preformed, but tran-
sient structure in the linker region and the flexible nat-
ure of domains 1 and 2 for the efficient functioning of
this protein [122]. Figure 3B shows that p27
Kip1
is pre-
dicted to be mostly disordered by both PONDRÒ VL-
XT and VSL1. Importantly, a region of predicted order
overlaps with the fragment of p27
Kip1
shown to contain
a significant amount of regular secondary structure in
its complex with Cdk2–cyclin A.
The above-mentioned sequential folding-upon-bind-
ing mechanism has been suggested to be crucial for the
selective inhibition of specific Cdk–cyclin complexes by
corresponding CKIs. Furthermore, p21
Waf1 ⁄ Cip1 ⁄ Sdi1
and p27
Kip1
target the cell cycle CDKs (Cdk1, Cdk2,
Cdk3, Cdk4 and Cdk6) but fail to bind and inhibit

X-ray structure determination, NMR analyses revealed
that residues 77–81 in the middle of the central helix
were highly flexible and functioned as a hinge [136].
This hinge facilitates a binding mode in which CaM
surrounds the target regions of its partners within the
two Ca
2+
-binding, globular domains, and in some
cases the hinge region remains unstructured after
association with the CaM target [132].
The interior faces of the globular domains have
features that accommodate target diversity, such as
nonrigid helix–helix packing that allow backbone
adjustments and high methionine contents that are
especially adept at side chain adjustments [132]. An
important structural feature enabling intermolecular
binding with maximal surface area of interaction is the
flexible connector between the two globular domains.
This flexibility accommodates a high diversity of
sequence features in the target by allowing the CaM
surface to seek complementary interactions by samp-
ling different positions and orientations relative to the
binding target surface. Additionally, the flexible hinge
facilitates variable separation of the two globular
domains after binding has occurred, again allowing for
binding diversity [136]. Despite the small size of the
disordered region (just five residues) and the slight
Flexible nets A. K. Dunker et al.
5138 FEBS Journal 272 (2005) 5129–5148 ª 2005 FEBS
amount of disorder in the entire protein (just 3%), this

cyclin A (A) magenta. In these panels, the
subunit not present in the experimental
binding reactions is gray to emphasize the
relevance of experimental data for binary
binding reactions to the mechanism of bind-
ing the Cdk2–cyclin A complex (right).
(Modified from [122] with the permission of
the authors, ª Nature Publishing Group,
1998–2005.) (B) PONDRÒ VL-XT (red) and
VSL1 (dark pink) predictions on the human
p27
Kip1
sequence (Swiss-Prot accession
number P46527). Green horizontal bars cor-
respond to the regions of the protein
involved in interaction with Cdk2 and
cyclin A: 1, domain 1 interacting with
cyclin A; 2, a linker helix involved in binding
both cyclin A and Cdk2; 3, domain 2 inter-
acting with Cdk2 [122]. Blue and dark yellow
horizontal bars correspond to the helices
and b-structure stabilized by the formation
of a triple complex, cyclin A–Cdk2–p27
Kip1
.
A. K. Dunker et al. Flexible nets
FEBS Journal 272 (2005) 5129–5148 ª 2005 FEBS 5139
because association-induced local water removal
increases this hydrogen-bond stability [141]. Important
for the present work, the application of the dehydron

but no evidence has yet been presented that dehydrons
have a similar capability. Thus, mutation-driven vari-
ation of locally disordered regions is more likely than
dehydrons alone to be one of the key structural factors
leading to the evolution of hub proteins.
How do ID regions work?
To explain one of the potential mechanisms used by ID
proteins to interact with their binding partners, the con-
cept of ‘molecular recognition element’ (MoRE) was
introduced [145,146]. The MoRE defines an interaction-
prone short segment of disorder that becomes ordered
upon specific binding. There are three basic types of
MoREs: those that form a-helical structures upon bind-
ing; those that form b-strands (in which the peptide
forms a b-sheet with additional b-strands provided by
the protein partner); and those that form irregular struc-
tures when binding [48,145–147]. Proposed names for
these structures are a-MoRE, b-MoRE, and I-MoRE,
respectively [48]. Of course, a given MoRE could be
composed of more than one type of secondary structure
type when bound to its partner, resulting in complex
MoREs, such as I-a-I-MoREs, a-b-MoREs, etc.
MoREs can be detected experimentally as segments
of disordered regions that maintain some residual
structure (as in the case of p27
Kip1
[122]). They also
can be discovered by analysis of protein–protein com-
plexes deposited in the Protein Data Bank (PDB) [148]
that consist of short nonglobular protein fragments

[146,147]. This study suggests that disorder-to-helix
transitions are common in protein interaction networks,
but laboratory experiments are needed to test whether
this mechanism is indeed as common as suggested.
Concluding remarks
Systematic postgenome proteome-wide analysis of pro-
tein interactions using large-scale two-hybrid screens
suggest that these interactions can be described as
complex SFNs [9,26,151]. On the other hand, many
traditional approaches have been developed to analyze
interactions, coordination, signaling and regulation on
Flexible nets A. K. Dunker et al.
5140 FEBS Journal 272 (2005) 5129–5148 ª 2005 FEBS
a smaller scale (e.g. within the scope of a single or
multiple interacting pathways). Quite often these
small-scale approaches yielded interesting results that
were used to develop models that correctly predicted
outcomes of changes and interruptions within the sys-
tem studied. Integrating small-scale network informa-
tion with global network information represents an
important but difficult task. Successful completion of
this task could lead to improved quality of the overall
data, shed light on the mechanisms of timing and regu-
lation, indicate how global and local properties of
complex macromolecular networks affect observable
biological properties (phenotype) and functions (physi-
ology) and suggest how changes in such properties
contribute to human diseases. Several groups are pro-
ductively working on the task of understanding inter-
esting subnetworks [152,153]. The example of GSK3b,

Studies along these lines are currently in progress.
Acknowledgements
We thank A.L. Barabasi for permission to modify
published figures and R.W. Kriwacki for assistance
with figure modification. We also thank Tanguy Le
Gall and Molecular Kinetics, Inc. for the use of the
proprietary fragment-mapping program to mine PDB
for some of the data presented in Table 1. The Indiana
Genomics Initiative (INGEN) and NIH Grant 1 R01
LM007688-0A1 provided support to A.K.D. This
work was supported in part by INTAS 2001-2347
Grant to V.N.U., and L.M.I. was supported by the
grant MCB-0444818 from the National Science Foun-
dation.
References
1 Goh KI, Oh E, Jeong H, Kahng B & Kim D (2002)
Classification of scale-free networks. Proc Natl Acad
Sci USA 99, 12583–12588.
2 Jeong H, Tombor B, Albert R, Oltvai ZN & Barabasi
AL (2000) The large-scale organization of metabolic
networks. Nature 407 , 651–654.
3 Barabasi AL & Albert R (1999) Emergence of scaling
in random networks. Science 286, 509–512.
4 Watts DJ & Strogatz SH (1998) Collective dynamics of
‘small-world’ networks. Nature 393, 440–442.
5 Barabasi AL & Bonabeau E (2003) Scale-free net-
works. Sci Am 288, 60–69.
6 Erdo
¨
sP&Re

A. K. Dunker et al. Flexible nets
FEBS Journal 272 (2005) 5129–5148 ª 2005 FEBS 5141
14 Han JD, Dupuy D, Bertin N, Cusick ME & Vidal M
(2005) Effect of sampling on topology predictions of
protein–protein interaction networks. Nat Biotechnol
23, 839–844.
15 Bork P, Jensen LJ, von Mering C, Ramani AK, Lee I
& Marcotte EM (2004) Protein interaction networks
from yeast to human. Curr Opin Struct Biol 14, 292–
299.
16 Cesareni G, Ceol A, Gavrila C, Palazzi LM, Persico
M & Schneider MV (2005) Comparative interactomics.
FEBS Lett 579, 1828–1833.
17 von Mering C, Krause R, Snel B, Cornell M, Oliver
SG, Fields S & Bork P (2002) Comparative assessment
of large-scale data sets of protein–protein interactions.
Nature 417, 399–403.
18 Bader GD & Hogue CW (2002) Analyzing yeast pro-
tein–protein interaction data obtained from different
sources. Nat Biotechnol 20, 991–997.
19 Hoffmann R & Valencia A (2003) Protein interaction:
same network, different hubs. Trends Genet 19, 681–
683.
20 Wu CH, Huang H, Nikolskaya A, Hu Z & Barker
WC (2004) The iProClass integrated database for
protein functional analysis. Comput Biol Chem 28,
87–96.
21 Huang TW, Tien AC, Huang WS, Lee YC, Peng CL,
Tseng HH, Kao CY & Huang CY (2004) POINT: a
database for the prediction of protein–protein interac-

30 Hasty J & Collins JJ (2001) Protein interactions.
Unspinning the web. Nature 411, 30–31.
31 Gunasekaran K, Tsai CJ, Kumar S, Zanuy D & Nussi-
nov R (2003) Extended disordered proteins: targeting
function with less scaffold. Trends Biochem Sci 28,
81–85.
32 Dunker AK, Brown CJ, Lawson JD, Iakoucheva LM
& Obradovic Z (2002) Intrinsic disorder and protein
function. Biochemistry 41, 6573–6582.
33 Dunker AK, Brown CJ & Obradovic Z (2002) Identifi-
cation and functions of usefully disordered proteins.
Adv Protein Chem 62, 25–49.
34 Dunker AK, Lawson JD, Brown CJ, Williams RM,
Romero P, Oh JS, Oldfield CJ, Campen AM, Ratliff
CM, Hipps KW et al. (2001) Intrinsically disordered
protein. J Mol Graph Model 19, 26–59.
35 Iakoucheva LM, Brown CJ, Lawson JD, Obradovic Z
& Dunker AK (2002) Intrinsic disorder in cell-signal-
ing and cancer-associated proteins. J Mol Biol 323,
573–584.
36 Uversky VN (2003) Protein folding revisited. A poly-
peptide chain at the folding-misfolding-nonfolding
cross-roads: which way to go? Cell Mol Life Sci 60,
1852–1871.
37 Uversky VN (2002) Natively unfolded proteins: a point
where biology waits for physics. Protein Sci 11, 739–
756.
38 Uversky VN (2002) What does it mean to be natively
unfolded? Eur J Biochem 269, 2–12.
39 Dyson HJ & Wright PE (2002) Coupling of folding

linking function and disorder. Nat Biotechnol 19,
805–806.
50 Uversky VN, Gillespie JR & Fink AL (2000) Why are
‘natively unfolded’ proteins unstructured under physio-
logic conditions? Proteins 41, 415–427.
51 Daughdrill GW, Pielak GJ, Uversky VN, Cortese MS
& Dunker AK (2005) Natively disordered proteins. In
Handbook of Protein Folding (Buchner J & Kiefhaber
T, eds), pp. 271–353. Wiley-VCH-Verlag GmbH
KGaA, Weinheim, Germany.
52 Romero P, Obradovic Z, Kissinger C, Villafranca JE
& Dunker AK (1997) Identifying disordered regions in
proteins from amino acid sequence. Proceedings of
International Conference on Neural Networks 1, 90–95.
53 Obradovic Z, Peng K, Vucetic S, Radivojac P, Brown
CJ & Dunker AK (2003) Predicting intrinsic disorder
from amino acid sequence. Proteins 53 (Suppl. 6), 566–
572.
54 Romero P, Obradovic Z & Dunker AK (1997)
Sequence data analysis for long disordered regions pre-
diction in the calcineurin family. Genome Inform Ser
Workshop Genome Inform 8, 110–124.
55 Romero P, Obradovic Z, Kissinger CR, Villafranca
JE, Garner E, Guilliot S & Dunker AK (1998) Thou-
sands of proteins likely to have long disordered
regions. Pac Symp Biocomput 437–448.
56 Dunker AK, Garner E, Guilliot S, Romero P, Albr-
echt K, Hart J, Obradovic Z, Kissinger C & Villa-
franca JE (1998) Protein disorder and the evolution of
molecular recognition: theory, predictions and observa-

(Balaban M, ed.), pp. 79–94. Elsevier ⁄ North-Holland
Biomedical Press, New York.
66 Kriwacki RW, Hengst L, Tennant L, Reed SI &
Wright PE (1996) Structural studies of p21Waf1 ⁄ -
Cip1 ⁄ Sdi1 in the free and Cdk2-bound state: conform-
ational disorder mediates binding diversity. Proc Natl
Acad Sci USA 93, 11504–11509.
67 Dajani R, Fraser E, Roe SM, Yeo M, Good VM,
Thompson V, Dale TC & Pearl LH (2003) Structural
basis for recruitment of glycogen synthase kinase 3beta
to the axin-APC scaffold complex. Embo J 22, 494–501.
68 Bax B, Carter PS, Lewis C, Guy AR, Bridges A, Tan-
ner R, Pettman G, Mannix C, Culbert AA, Brown
MJ, Smith DG & Reith AD (2001) The structure of
phosphorylated GSK-3beta complexed with a peptide,
FRATtide, that inhibits beta-catenin phosphorylation.
Structure (Camb) 9, 1143–1152.
69 Ding Y & Dale T (2002) Wnt signal transduction:
kinase cogs in a nano-machine? Trends Biochem Sci
27, 327–329.
70 Polakis P (2000) Wnt signaling and cancer. Genes Dev
14, 1837–1851.
71 He TC, Sparks AB, Rago C, Hermeking H, Zawel L,
da Costa LT, Morin PJ, Vogelstein B & Kinzler KW
(1998) Identification of c-MYC as a target of the APC
pathway. Science 281, 1509–1512.
72 Plyte SE, Hughes K, Nikolakaki E, Pulverer BJ &
Woodgett JR (1992) Glycogen synthase kinase-3: func-
tions in oncogenesis and development. Biochim Biophys
Acta 1114, 147–162.

80 Reeves R (2001) Molecular biology of HMGA pro-
teins: hubs of nuclear function. Gene 277, 63–81.
81 Lehn DA, Elton TS, Johnson KR & Reeves R (1988)
A conformational study of the sequence specific bind-
ing of HMG-I(Y) with the bovine interleukin-2 cDNA.
Biochem Int 16, 963–971.
82 Evans JN, Nissen MS & R (1992) Assignment of the
1H NMR spectrum of a consensus DNA-binding pep-
tide from the HMG-I protein. Bull Mag Reson 14,
171–174.
83 Evans JN, Zajicek J, Nissen MS, Munske G, Smith V
& Reeves R (1995) 1H and 13C NMR assignments
and molecular modelling of a minor groove DNA-
binding peptide from the HMG-I protein. Int J Pept
Protein Res 45 , 554–560.
84 Lund T, Holtlund J, Fredriksen M & Laland SG
(1983) On the presence of two new high mobility
group-like proteins in HeLa S3 cells. FEBS Lett 152,
163–167.
85 Obradovic Z, Peng K, Vucetic S, Radivojac P &
Dunker AK (2005) Exploiting heterogeneous sequence
properties improves prediction of protein disorder.
Proteins: Struct Func Bioinf in press.
86 Reeves R & Nissen MS (1990) The A.T-DNA-binding
domain of mammalian high mobility group I chromo-
somal proteins. A novel peptide motif for recognizing
DNA structure. J Biol Chem 265, 8573–8582.
87 Huth JR, Bewley CA, Nissen MS, Evans JN, Reeves
R, Gronenborn AM & Clore GM (1997) The solution
structure of an HMG-I(Y)-DNA complex defines a

95 Friedberg EC, Walker GC & Siede W (1995) DNA
Repair and Mutagenesis. ASM Press, Washington, DC.
96 Bootsma D, Kreamer KH, Cleaver JE & Hoeijmakers
JHJ (1998) Nucleotide excision repair: xeroderma
pigmentosum, Cocayne syndrome, and trichothio-
dystrophy. In The Genetic Basis of Human Cancer
(Vogelstein B & Kinzler KW, eds), pp. 245–274.
McGraw-Hill, New York.
97 de Boer J & Hoeijmakers JH (2000) Nucleotide exci-
sion repair and human syndromes. Carcinogenesis 21,
453–460.
98 Cleaver JE & States JC (1997) The DNA damage-
recognition problem in human and other eukaryotic
cells: the XPA damage binding protein. Biochem J 328,
1–12.
99 Jones CJ & Wood RD (1993) Preferential binding of
the xeroderma pigmentosum group A complementing
protein to damaged DNA. Biochemistry 32, 12096–
12104.
100 Asahina H, Kuraoka I, Shirakawa M, Morita EH,
Miura N, Miyamoto I, Ohtsuka E, Okada Y &
Tanaka K (1994) The XPA protein is a zinc metallo-
protein with an ability to recognize various kinds of
DNA damage. Mutat Res 315, 229–237.
101 Nitta M, Saijo M, Kodo N, Matsuda T, Nakatsu Y,
Tamai H & Tanaka K (2000) A novel cytoplasmic
GTPase XAB1 interacts with DNA repair protein
XPA. Nucleic Acids Res 28, 4212–4218.
102 Nakatsu Y, Asahina H, Citterio E, Rademakers S,
Vermeulen W, Kamiuchi S, Yeo JP, Khaw MC, Saijo

108 Park CH, Mu D, Reardon JT & Sancar A (1995) The
general transcription-repair factor TFIIH is recruited
to the excision repair complex by the XPA protein
independent of the TFIIE transcription factor. J Biol
Chem 270, 4896–4902.
109 Nocentini S, Coin F, Saijo M, Tanaka K & Egly JM
(1997) DNA damage recognition by XPA protein pro-
motes efficient recruitment of transcription factor II H.
J Biol Chem 272, 22991–22994.
110 Kuraoka I, Morita EH, Saijo M, Matsuda T, Mori-
kawa K, Shirakawa M & Tanaka K (1996) Identifica-
tion of a damaged-DNA binding domain of the XPA
protein. Mutat Res 362, 87–95.
111 Ikegami T, Kuraoka I, Saijo M, Kodo N, Kyogoku Y,
Morikawa K, Tanaka K & Shirakawa M (1998) Solu-
tion structure of the DNA- and RPA-binding domain
of the human repair factor XPA. Nat Struct Biol 5,
701–706.
112 Buchko GW, Ni S, Thrall BD & Kennedy MA (1998)
Structural features of the minimal DNA binding
domain (M98–F219) of human nucleotide excision
repair protein XPA. Nucleic Acids Res 26, 2779–2788.
113 Iakoucheva LM, Kimzey AL, Masselon CD, Bruce JE,
Garner EC, Brown CJ, Dunker AK, Smith RD &
Ackerman EJ (2001) Identification of intrinsic order
and disorder in the DNA repair protein XPA. Protein
Sci 10, 560–571.
114 Fontana A, de Laureto PP, Spolaore B, Frare E,
Picotti P & Zambonin M (2004) Probing protein struc-
ture by limited proteolysis. Acta Biochim Pol 51, 299–

cyclin-CDK complexes through a sequential mechan-
ism involving binding-induced protein folding. Nat
Struct Mol Biol 11, 358–364.
123 Harper JW, Adami GR, Wei N, Keyomarsi K &
Elledge SJ (1993) The p21 Cdk-interacting protein
Cip1 is a potent inhibitor of G1 cyclin-dependent
kinases. Cell 75, 805–816.
124 Harper JW, Elledge SJ, Keyomarsi K, Dynlacht B,
Tsai LH, Zhang P, Dobrowolski S, Bai C, Connell-
Crowley L, Swindell E, Fox MP & Wei N (1995) Inhi-
bition of cyclin-dependent kinases by p21. Mol Biol
Cell 6, 387–400.
125 Polyak K, Lee MH, Erdjument-Bromage H, Koff A,
Roberts JM, Tempst P & Massague J (1994) Cloning
of p27Kip1, a cyclin-dependent kinase inhibitor and a
potential mediator of extracellular antimitogenic sig-
nals. Cell 78, 59–66.
126 Toyoshima H & Hunter T (1994) p27, a novel inhibi-
tor of G1 cyclin-Cdk protein kinase activity, is related
to p21. Cell 78, 67–74.
127 Matsuoka S, Edwards MC, Bai C, Parker S, Zhang P,
Baldini A, Harper JW & Elledge SJ (1995) p57KIP2, a
structurally distinct member of the p21CIP1 Cdk inhi-
bitor family, is a candidate tumor suppressor gene.
Genes Dev 9, 650–662.
128 Flaugh SL & Lumb KJ (2001) Effects of macromole-
cular crowding on the intrinsically disordered proteins
c-Fos and p27(Kip1). Biomacromolecules 2, 538–540.
A. K. Dunker et al. Flexible nets
FEBS Journal 272 (2005) 5129–5148 ª 2005 FEBS 5145

138 Fernandez A, Colubri A & Berry RS (2002) Three-
body correlations in protein folding: the origin of
cooperativity. Physica A 307, 235–259.
139 Fernandez A & Scheraga HA (2003) Insufficiently
dehydrated hydrogen bonds as determinants of protein
interactions. Proc Natl Acad Sci USA 100, 113–118.
140 Fernandez A & Scott LR (2003) Adherence of packing
defects in soluble proteins. Phys Rev Lett 91, 018102.
141 Fernandez A & Scott R (2003) Dehydron: a structu-
rally encoded signal for protein interaction. Biophys J
85, 1914–1928.
142 Fernandez A, Scott R & Berry RS (2004) The noncon-
served wrapping of conserved protein folds reveals a
trend toward increasing connectivity in proteomic net-
works. Proc Natl Acad Sci USA 101, 2823–2827.
143 Qin H, Lu HH, Wu WB & Li WH (2003) Evolution of
the yeast protein interaction network. Proc Natl Acad
Sci USA 100, 12820–12824.
144 Fernandez A & Berry RS (2004) Molecular dimension
explored in evolution to promote proteomic complex-
ity. Proc Natl Acad Sci USA 101, 13460–13465.
145 Bourhis JM, Johansson K, Receveur-Brechot V, Old-
field CJ, Dunker KA, Canard B & Longhi S (2004)
The C-terminal domain of measles virus nucleoprotein
belongs to the class of intrinsically disordered proteins
that fold upon binding to their physiological partner.
Virus Res 99, 157–167.
146 Oldfield CJ (2002) Predictions of disordered proteins
and protein binding regions: applications to structural
genomics. Honors Thesis, Washington State University

Biochemistry 35, 13709–13715.
155 Dev KK, van der Putten H, Sommer B & Rovelli G
(2003) Part I: parkin-associated proteins and Parkin-
son’s disease. Neuropharmacology 45, 1–13.
156 Dev KK, Hofele K, Barbieri S, Buchman VL & van
der Putten H (2003) Part II: alpha-synuclein and its
molecular pathophysiological role in neurodegenerative
disease. Neuropharmacology 45, 14–44.
157 Lynch WP, Riseman VM & Bretscher A (1987)
Smooth muscle caldesmon is an extended flexible
monomeric protein in solution that can readily
undergo reversible intra- and intermolecular sulfhydryl
cross-linking. A mechanism for caldesmon’s F-actin
bundling activity. J Biol Chem 262, 7429–7437.
158 Gerthoffer WT (2005) Signal-transduction pathways
that regulate visceral smooth muscle function III. Coup-
ling of muscarinic receptors to signaling kinases and
effector proteins in gastrointestinal smooth muscles. Am
J Physiol Gastrointest Liver Physiol 288, G849–G853.
159 Polyakov AA, Huber PA, Marston SB & Gusev NB
(1998) Interaction of isoforms of S100 protein with
smooth muscle caldesmon. FEBS Lett 422, 235–239.
Flexible nets A. K. Dunker et al.
5146 FEBS Journal 272 (2005) 5129–5148 ª 2005 FEBS
160 Huber PA (1997) Caldesmon. Int J Biochem Cell Biol
29, 1047–1051.
161 Reeves R & Nissen MS (1999) Purification and assays
for high mobility group HMG-I(Y) protein function.
Methods Enzymol 304, 155–188.
162 Sgarra R, Tessari MA, Di Bernardo J, Rustighi A,

VAMP-binding protein, SNAP-23, expressed in non-
neuronal tissues. J Biol Chem 271, 13300–13303.
169 Mark WY, Liao JC, Lu Y, Ayed A, Laister R, Szymc-
zyna B, Chakrabartty A & Arrowsmith CH (2005)
Characterization of segments from the central region
of BRCA1: an intrinsically disordered scaffold for mul-
tiple protein–protein and protein–DNA interactions?
J Mol Biol 345, 275–287.
170 Yoshida K & Miki Y (2004) Role of BRCA1 and
BRCA2 as regulators of DNA repair, transcription,
and cell cycle in response to DNA damage. Cancer Sci
95, 866–871.
171 Saijo M, Matsuda T, Kuraoka I & Tanaka K (2004)
Inhibition of nucleotide excision repair by anti-XPA
monoclonal antibodies which interfere with binding to
RPA, ERCC1, and TFIIH. Biochem Biophys Res
Commun 321, 815–822.
172 Warnmark A, Wikstrom A, Wright AP, Gustafsson
JA & Hard T (2001) The N-terminal regions of
estrogen receptor alpha and beta are unstructured
in vitro and show different TBP binding properties.
J Biol Chem 276, 45939–45944.
173 Liu G, Schwartz JA & Brooks SC (2000) Estrogen
receptor protects p53 from deactivation by human
double minute-2. Cancer Res 60, 1810–1814.
174 Fan S, Wang J, Yuan R, Ma Y, Meng Q, Erdos MR,
Pestell RG, Yuan F, Auborn KJ, Goldberg ID &
Rosen EM (1999) BRCA1 inhibition of estrogen recep-
tor signaling in transfected cells. Science 284, 1354–
1356.

J, Levine AJ & Pavletich NP (1996) Structure of the
MDM2 oncoprotein bound to the p53 tumor suppressor
transactivation domain. Science 274, 948–953.
183 Bothner B, Lewis WS, DiGiammarino EL, Weber JD,
Bothner SJ & Kriwacki RW (2001) Defining the mole-
cular basis of Arf and Hdm2 interactions. J Mol Biol
314, 263–277.
184 Iwakuma T & Lozano G (2003) MDM2, an introduc-
tion. Mol Cancer Res 1, 993–1000.
185 Kissinger CR, Parge HE, Knighton DR, Lewis CT,
Pelletier LA, Tempczyk A, Kalish VJ, Tucker KD,
Showalter RE, Moomaw EW, Gastinel LN, Habuka
N, Chen X, Maldonado F, Barker JE, Bacquet R &
Villafranca JE (1995) Crystal structures of human
A. K. Dunker et al. Flexible nets
FEBS Journal 272 (2005) 5129–5148 ª 2005 FEBS 5147
calcineurin and the human FKBP12-FK506-calcineurin
complex. Nature 378 , 641–644.
186 Im SH & Rao A (2004) Activation and deactivation of
gene expression by Ca2+ ⁄ calcineurin-NFAT-mediated
signaling. Mol Cells 18, 1–9.
187 Genesca L, Aubareda A, Fuentes JJ, Estivill X, De La
Luna S & Perez-Riba M (2003) Phosphorylation of
calcipressin 1 increases its ability to inhibit calcineurin
and decreases calcipressin half-life. Biochem J 374,
567–575.
188 Sun L, Youn HD, Loh C, Stolow M, He W & Liu JO
(1998) Cabin 1, a negative regulator for calcineurin sig-
naling in T lymphocytes. Immunity 8, 703–711.
189 Banerjee A, Banks AS, Nawijn MC, Chen XP & Roth-

197 Ballweber E, Galla M, Aktories K, Yeoh S, Weeds
AG & Mannherz HG (2001) Interaction of ADP-ribo-
sylated actin with actin binding proteins. FEBS Lett
508, 131–135.
198 Xia Z & Storm DR (2005) The role of calmodulin as
a signal integrator for synaptic plasticity. Nat Rev
Neurosci 6, 267–276.
199 Takahashi K, Hiwada K & Kokubu T (1988) Vascular
smooth muscle calponin. A novel troponin T-like pro-
tein. Hypertension 11, 620–626.
200 Gusev NB (2001) Some properties of caldesmon and
calponin and the participation of these proteins in
regulation of smooth muscle contraction and cytoskele-
ton formation. Biochemistry (Mosc) 66, 1112–1121.
201 Bader GD, Betel D & Hogue CW (2003) BIND: the
Biomolecular Interaction Network Database. Nucleic
Acids Res 31, 248–250.
202 Salwinski L, Miller CS, Smith AJ, Pettit FK, Bowie JU
& Eisenberg D (2004) The Database of Interacting Pro-
teins: 2004 update. Nucleic Acids Res 32, D449–D451.
203 von Mering C, Jensen LJ, Snel B, Hooper SD, Krupp
M, Foglierini M, Jouffre N, Huynen MA & Bork P
(2005) STRING: known and predicted protein–protein
associations, integrated and transferred across organ-
isms. Nucleic Acids Res 33, D433–D437.
204 Peri S, Navarro JD, Kristiansen TZ, Amanchy R,
Surendranath V, Muthusamy B, Gandhi TK, Chandri-
ka KN, Deshpande N, Suresh S et al. (2004) Human
protein reference database as a discovery resource for
proteomics. Nucleic Acids Res 32, D497–D501.