You are on page 1of 21

1478 DOI 10.1002/pmic.

201100563 Proteomics 2012, 12, 1478–1498

REVIEW

History of protein–protein interactions: From egg-white


to complex networks
Pascal Braun1,2 and Anne-Claude Gingras3,4
1
Department of Plant Systems Biology, Center for Life and Food Sciences Weihenstephan, Technical University
Munich, Freising, Germany
2
Research Unit Protein Science, Helmholtz Centre Munich, Munich, Germany
3
Samuel Lunenfeld Research Institute at Mount Sinai Hospital, Toronto, Ontario, Canada
4
Department of Molecular Genetics, University of Toronto, Toronto, Ontario, Canada

Today, it is widely appreciated that protein–protein interactions play a fundamental role in Received: October 27, 2011
biological processes. This was not always the case. The study of protein interactions started Revised: January 4, 2012
Accepted: January 17, 2012
slowly and evolved considerably, together with conceptual and technological progress in differ-
ent areas of research through the late 19th and the 20th centuries. In this review, we present
some of the key experiments that have introduced major conceptual advances in biochemistry
and molecular biology, and review technological breakthroughs that have paved the way for
today’s systems-wide approaches to protein–protein interaction analysis.

Keywords:
Biochemistry / History / Networks / Protein–protein interactions / Signaling pathways
/ Systems biology

1 Introduction into physical motion, and maintaining cellular organiza-


tion. In recent years, it has become apparent that proteins
Proteins play central roles in all aspects of cellular func- form intricate networks via their interactions with other pro-
tion including metabolism, information processing, deci- teins, resulting in highly organized, dynamic cellular sys-
sion making, transport, and structural organization. Pro- tems [1]. Analysis of such interactome networks has provided
teins mediate their functions by physically interacting with important insights into the global organization of cellular
proteins and other molecules such as metabolites, lipids, systems.
and nucleic acids. Protein–protein interactions play key Because of the critical importance of protein interactions,
roles for mediating functions such as sensing the environ- when an unknown protein is investigated today, one of the
ment, mediating signal transduction, adjusting the activ- first questions asked is: what other proteins does it inter-
ity of metabolic and signaling enzymes, converting energy act with? This was not always the case. The study of protein
interactions started slowly and evolved considerably, hand-in-
hand with conceptual and technological progress in different
Correspondence: Dr. Pascal Braun, Department of Plant Sys- areas of biological research (Fig. 1). In this process, no single
tems Biology, Center for Life and Food Sciences Weihenstephan, point can be identified at which the importance of physical
Technische Universität München, 85354 Freising, Germany and protein–protein interactions was suddenly realized. Instead,
Helmholtz Centre Munich, Research Unit Protein Science, In- again and again physical interactions were found to play key
golstädter Landstraße 1 85764 München/Oberschleißheim, Ger- roles in processes that eventually included essentially all bi-
many
ological functions. We think it is timely to revisit how we
E-mail: pbraun@wzw.tum.de
Fax: +49-8161-71-2886 came to recognize the diverse roles exerted by protein–protein
interactions. In this narrative, we present some of the key ex-
Abbreviations: AP-MS, affinity purification coupled to mass spec- periments that have introduced major conceptual changes
trometry; HT, high-throughput; IP, immunoprecipitation; MPF,
maturation promoting factor; PKA, protein kinase A; SV40, simian
vacuolating virus 40; Y2H, yeast-two-hybrid Colour Online: See the article online to view Figs. 1 and 4 in colour.


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
Proteomics 2012, 12, 1478–1498 1479

Figure 1. Timeline of protein-protein interaction research. In the top part conceptual advances and discoveries are indicated, in the lower
part technological advances and inventions are indicated.

in biochemistry and molecular biology, and review techno- first such activity, diastase, which catalyzes the conversion of
logical breakthroughs that have paved the way for today’s starch to sugar, was discovered in 1833 in an alcohol precip-
systems-wide approaches to protein interaction analysis. We itate of malt extract by French chemists working in a sugar
apologize for the unavoidable omission of findings due to factory [5]. Soon after, a digestive activity in animal stomach
space limitations. was discovered and named pepsin (1835) [6]. In 1876, Kühne
proposed the term “enzyme” for such ferments that were ac-
tive independent of an organism [6]. In the same article, he
2 Proteins, enzymes, and first protein described the isolation of trypsin – the digestive enzyme of the
interactions pancreas [7], which became a member of one of the earliest
protein interaction pairs. In 1906, a trypsin inhibiting activity
The first protein interactions were discovered even before was detected in the albumin fraction of blood serum [8]. In a
there was a clear understanding on what proteins are. The quantitative kinetic study, the nature of this interaction was
term “protein” (gr. proteios–of the first kind) was proposed analyzed. Assuming an interaction according to the law of
in 1838 by Berzelius in a letter to the Dutch chemist, Mulder. mass action, i.e. two soluble molecules, the activity of trypsin
Mulder had found the same chemical formula for organic was quantitatively predicted for several experiments (Fig. 2).
material from plants and animals (C400 H620 N100 O120 P1 S1 ), Importantly, the rate at which the enzyme activity decreased
which led him and Berzelius to the assumption that this when either pure enzyme or enzyme-inhibitor complex was
organic material was made of a single type of molecule [2]. diluted was slower for the latter. This could be explained by
At this point, it was unclear, however, how this substance dissociation of the enzyme-inhibitor complex and this result
related to biological processes. In the same decade, it was led to the conclusion that in contrast to inhibition by absorp-
shown that alcoholic fermentation was caused by yeast. Louis tion to charcoal, the inhibitor and the enzyme “combine to
Pasteur later described that different types of fermentation form an inactive but dissociable compound” [9]. This interac-
all require intact cells and considered fermentation a process tion between the enzyme trypsin and its inhibitor is thus one
irreducibly associated with life [3]. This was often taken as of the first regulatory protein–protein interactions identified.
evidence for the view that living organisms contain a vital Despite these early insights, it took well into the 20th century
force that was distinct from physicochemical processes (“vi- until it was convincingly shown that enzymes are proteins.
talism”) [4]. However, biological activities that acted outside of In 1926, Sumner purified the major protein in a preparation
organisms had already been identified in the early 1830s. The of the enzyme urease such that it could be crystallized. Upon


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
1480 P. Braun and A.-C. Gingras Proteomics 2012, 12, 1478–1498

Figure 2. Complex formation between


trypsin and trypsin inhibitor. (A) Trypsin
activity is measured by the change in con-
ductivity of a gelatin (substrate) solution
caused by trypsin activity during a de-
fined time interval. (B) Enzyme activity in
different conditions is predicted assuming
a reversible equilibrium between trypsin
and its inhibitor according to the law of
mass action. Calculated and observed en-
zymatic activities for dilutions of trypsin (I)
and trypsin with inhibitor (II) by the indi-
cated volume of water. (Original data from
Northrop, 1922 [9])

dissolving the protein crystals in water, the solution exhib- living cells. This ultimately disproved vitalism and set the
ited very high enzymatic activity thus demonstrating that the foundation for modern biochemistry. By the turn of the cen-
crystallized protein was identical to the enzyme [10]. tury, enough evidence had accumulated to suggest that en-
Another of the earliest protein–protein interactions was zymes could be “the chemical basis of life” [15]. With this,
detected in the context of investigating infectious diseases: understanding the molecules and chemical processes of life
the interaction between antibodies and antigens. In the sec- became essential for biology. Although physical interactions
ond half of the 19th century, it was known that some infec- among proteins had been identified, this did not immediately
tious diseases are caused by toxins secreted by pathogens. In trigger intense research in their functions. First, the molecu-
the absence of sufficient quantities of pathogen toxins, many lar nature of proteins needed to be elucidated which became
scientists explored the physiological response of animals to a central topic in the new field of biochemistry.
plant toxins. In 1891, Paul Ehrlich published results on the
closely related and highly toxic plant proteins, ricin and abrin
[11]. In the first study, he reported the ability to immunize 3 The nature of proteins
mice against the toxic effects of ricin. He explained this by
the formation of an antitoxin, antiricin, in the blood of the For a long time, the nature of proteins and their relation-
animals. The existence of antiricin was demonstrated by tem- ship to enzymes was unclear. Due to their molecular weight,
porarily transferring immunity to naı̈ve animals by injecting which is orders of magnitude larger than any then known
them with antiricin containing blood. In the second article, he molecule, many scientists doubted that proteins could be in-
conducted a detailed toxicological study of the closely related dividual molecules [16–18]. In the 1910s and throughout the
abrin. Both ricin and abrin led to similar toxicologic pheno- 1920s, it was hotly debated whether proteins might be col-
types and had been proposed to be identical. In his studies, loids: heterogeneous, noncovalent compound assemblies of
Ehrlich found a number of differences between the two sub- smaller molecules and electrolytes [16–18]. Toward the end of
stances to support the conclusion that they were likely differ- the 1920s, significant progress was made in elucidating the
ent. As with ricin, however, he found that it was possible to nature of proteins and their relation to enzymes. One impor-
immunize animals against abrin by feeding nontoxic doses, tant finding was that proteins can be purified and crystallized,
resulting in the appearance of antiabrin in the blood of the which was the requirement for Sumner’s demonstration that
animals. Crucially, despite extensive pathological, and, as we enzymes are proteins (see above) [10]. The invention of ul-
know today, extensive structural similarities, animals that are tracentrifugation by Svedberg contributed decisive evidence
immune against one toxin are still fully sensitive toward the that proteins were in fact large molecules with defined chem-
other. Ehrlich introduced the term “antikörper” (antibodies) ical properties. In contrast to the diffuse appearance of even
to define the class of substances with the antitoxic activities the most homogeneous colloids, proteins yielded a single ho-
that are produced by the body. Mirroring today’s widespread mogeneous border in these experiments (Fig. 3A) suggesting
use as a technological tool for identification and highly spe- that “notwithstanding their size and enormous weight, the hemo-
cific purification, Ehrlich used antibodies as a tool to establish cyanin particles are entitled to be considered as real molecules
identity between two proteins “if two substances give rise to two acting as single units” [19, 20]. During his centrifugation ex-
different antikörper, then they themselves must be different” [12]. periments, Svedberg also noticed that several sharp bands
It took another 30 years, however, until it was shown that became visible at higher pH, corresponding to proteins of
antibodies are in fact proteins [13]. smaller molecular weight (Fig. 3B). Upon detailed analysis
Shortly after Ehrlich’s discovery of antibodies, in 1897, of hemocyanins from several species, he realized that the
Buchner demonstrated that fermentation of sugar can take molecular weights of large species always corresponded to
place in extracts from yeast lysate [14], i.e. in the absence of multiples of the smallest detectable protein, suggesting that


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
Proteomics 2012, 12, 1478–1498 1481

A the development of methods to label the proteins after


partial acid hydrolysis (and in some case cleavage by pro-
teases), fractionation of the products, followed by end-group
identification [24–30]. This was a tedious method, with no au-
tomation, random cleavage of the protein chain, and manual
assembly of protein sequences from small peptide sequences
identified. While the approach could not easily be applied
to larger proteins, the sequencing of insulin unambiguously
demonstrated the noncolloidal, unique, and nonrepetitive
nature of protein primary structure, which Sanger himself
summarized in a 1952 review on this topic: “It has frequently
been suggested that proteins may not be pure chemical entities
B but may consist of mixtures of closely related substances with no
absolute unique structure. The chemical results obtained so far
suggest that this is not the case, and that a protein is really a single
chemical substance, each molecule of one protein being identical
to every other molecule of the same pure protein” [31]. In the fol-
lowing year, Sanger and Thompson completed the sequence
of the insulin A chain, which they found to be different from
that of the B chain, leading them to conclude: “It would thus
seem [ . . . ] that each protein has its own unique (amino acid)
arrangement; an arrangement that endows it with its particular
Figure 3. Demonstration that hemocyanins are molecules and
consist of subunits. Schematic representation of photographic
properties and specificities and fits it for the function that it per-
plates of ultracentrifuge sedimentation experiments. Sedimen- forms in nature.” [32,33]. The chemical nature of proteins was
tation is performed at constant centrifugal force, picture series settled.
represent 5 min intervals, except bottom left 10 min. (A) Sedi- Once proteins were shown to consist of amino acids
mentation of hemocyanins (top) and gold colloids (bottom). Gold that are arranged in a specific sequence in each protein, an
colloids, among the most homogeneous colloids known, sedi- important question became how the different amino acids
ment with a diffuse appearance. In contrast, proteins exhibit a
chains were arranged in three-dimensional structures to yield
sharp boundary suggesting they are a single type of molecule.
(B) Sedimentation of hemocyanins at lower pH (top) and higher molecules that have biological activity. Another monumental
pH (bottom). Original experiments were done with hemocyanins step enabled such structural analysis of pure protein crystals
from different species in [22]. At lower pH a single sharp sedi- subjected to x-rays: the conceptual solution of the “phase prob-
mentation front is observed. When the pH is increased (bottom) lem” by Max Perutz in the early 1950s (see also [34]). Together
several sharp fronts corresponding to protein species of lower with the arrival of computer technology, this enabled solving
molecular weights, later shown to be subunits, become visible.
the three dimensional structures of myosin and hemoglobin
(Data adapted from Svedberg, 1929 [20]; and Inga et al., 1936 [21])
in 1959 and 1960, respectively [35, 36]. These structures illus-
trated visually how the primary sequence of proteins folds into
a unique three-dimensional shape. Moreover, in the struc-
small subunits aggregate to form larger protein complexes. ture of hemoglobin, the oxygen-carrying protein of red blood
Moreover, the apparent dissociation was reversible [19, 21]. cells, the tetrameric subunit composition of the holoprotein
Without a conceptual distinction between peptide chains and was apparent, which provided an atomic perspective into the
proteins at that time, he hypothesized that different proteins quaternary arrangements of polypeptide chains in a protein
might be assembled from few elementary subunits of approx- complex and thus provided an explanation for Svedberg’s
imately 34kDa, which he proposed as a “general plan” about observation of the subunit dissociation of the evolutionary
the nature of proteins [20, 21]. In hindsight, these results are related hemocyanins. Conceptually, the structures laid the
among the first examples demonstrating that proteins can groundwork for understanding of allosteric interactions [37],
exist as dissociable complexes in vivo and mark the discovery and recognition of specific interaction interfaces [36, 38].
of quaternary structure [22]. With these landmark papers, together with publication of the
If proteins were in fact macromolecules of unprecedented double helix a few years earlier [39], the world of biochemistry
size, questions about their composition and assembly acquired a new dimension of visual accessibility.
became pressing. In the absence of a mechanism by which
highly complex peptide sequences could reproducibly be 4 Interactions give structure and regulate
synthesized by the cell, protein models in which peptides enzymes
are arranged in a repetitive manner seemed reasonable [23].
In the late 1940s, Frederick Sanger was able to determine Despite the early identification of interactions between
the sequence of the insulin B chain. This had required trypsin and its inhibitor and that of antibodies with their


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
1482 P. Braun and A.-C. Gingras Proteomics 2012, 12, 1478–1498

antigens, investigation of physical interactions between pro- tures mediate a wide range of biological functions related to
teins played a minor role throughout the early 20th century. both structural organization and motility including cellular
Starting in the 1940s, however, more and more examples motion, intracellular cargo transport, and exertion and sens-
of protein–protein interactions were uncovered, and the im- ing of mechanical forces and are a topic of intense research
portance of physical associations between proteins started to to this day [59].
become appreciated.

4.2 Protein interactions regulate metabolism


4.1 Physical motion mediated by interactions
between myosin and actin In the mid 1940s, the understanding of metabolism was rev-
olutionized by conceptual and technical advances. In 1944,
Physical motion has intrigued humans since antiquity, when Beadle and Tatum published the one-gene-one-enzyme hy-
Aristotle first described how muscles cause motion [40]. In pothesis, and demonstrated how genetic screens can be used
addition to electrophysiological studies on intact muscle [41], to identify new enzymatic reactions by isolation of mutants
myosin was identified as a major molecular constituent of that lost the ability to grow on media lacking specific metabo-
muscles by Kühne in the middle of the 19th century. In the lites [60]. Although initially met with resistance and deemed
1930s, studies of myosin revealed its elongated shape [42] irrelevant by some of the most eminent geneticists at the time,
and ATPase activity [43–45]. Importantly, the conformation their results demonstrated that complex processes such as
of myosin, measured by its viscosity, was shown to reversibly metabolism were amenable to genetic analysis [61]. In subse-
change upon addition of ATP, which suggested “for the first quent years, genes encoding enzymes in numerous biochem-
time” that myosin was a “contractible enzyme” [46]. In the fol- ical pathways were identified. At the same time, an increasing
lowing years, it became apparent that one of the previously number of metabolic enzymes was purified and investigated
described forms of myosin, “Myosin B”, was likely a com- in biochemical detail. For many of these, molecular weights
plex of actin and myosin, which physically interact to yield in the megadalton range were observed, which was very large
actomyosin [47–49] (Fig. 4). The physical interaction between even for proteins. For example, purified ␣-ketoglutaric oxi-
F-actin and myosin was later directly demonstrated in im- dase was found to sediment with a molecular weight of two
munoprecipitation (IP) experiments [50]. Moreover, it was megadaltons, suggesting that “[ . . . ] large molecules of pyruvic
shown that the interaction with actin triggers the ATPase and ␣-ketoglutaric oxidase may still be complexes of more than one
activity of myosin and thus, that the physical interaction be- protein. If so, they would represent organized units below the mito-
tween actin and myosin is critical for enzymatic activity [47]. chondrial level” [62]. Despite this early and correct speculation,
Thus, “the modern era (of understanding motion) began with it took nearly another 10 years for this enzyme complex to be
the demonstration that contraction is the result of the interaction resolved into its constituent protein components. Once the
of two proteins, actin and myosin with ATP [ . . . ]” [51]. individual proteins were identified, each could be associated
Actin, the second major protein of muscles, had been iden- with individual reaction steps of the overall catalyzed reac-
tified in 1887 by its ability to coagulate myosin. However, only tion [63]. Notably, in reassociation experiments described in
after the realization that Myosin B was a complex of two pro- the same paper, the original protein complex could be recon-
teins, actin was purified, named, and characterized in 1942. It stituted from the individual proteins, clearly demonstrating
was found to exist in two distinct forms, filamentous (F-) and the noncovalent nature of the interactions. Studies on other
globular (G-) actin [51–54]. In 1943, it was demonstrated that enzymes came to similar conclusions [64]. Over the next 10
G-actin could be obtained by desalting and thus depolymer- years, facilitated by progress in gel electrophoresis, the com-
izing F-actin [53]. The immunologic identity between both position of more and more enzymes was determined. In light
forms was further supported by the mutual recognition of of “hundreds of reports” describing the molecular weights of
each actin species by antibodies targeting the other form enzymes and their subunits, in 1971 it was noted that “it
[50]. In 1950, Straub demonstrated that removal of ATP from would appear that the vast majority of proteins are composed of
G-actin prevented polymerization into F-actin [54–56], and subunits” [65].
thereby, smooth muscle actin polymerization became an early The widespread identification of multimeric enzymes gen-
example of an ATP-driven protein–protein interaction. Over erated an interest in the functional aspects of these subunit
the next years, additional interactors for actin and myosin interactions, and evidence for subunit regulation accumu-
were identified, and the mechanism by which physically in- lated. In 1965, Monod and colleagues developed a model of
teracting proteins mediate physical motion was elucidated allosteric enzyme regulation, i.e. functional interactions of
[51, 57, 58]. Conceptually, actin fibers were also the first ex- different ligands that bind to spatially separated, i.e. non-
ample for cytoskeletal structures. In the following decades, overlapping binding sites. A central element of the model is
the ubiquitous role of physical protein–protein interactions symmetry of different binding sites for a given ligand as of-
in creating diverse subcellular structures became apparent ten found in protein complexes. At least two conformational
following the discoveries of microtubules and intermediate states of the oligomeric protein complex are postulated; tran-
filaments, as well as the nuclear pore complex. These struc- sition from one conformational state into the other results


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
Proteomics 2012, 12, 1478–1498 1483

Figure 4. Experimental demonstration that


interaction between myosin and actin un-
derlies physical motion. (A) Ostwald vis-
cosimeter used to determine the viscosity
of myosin solutions. Liquid is drawn to the
upper mark and released. From the time
required for the liquid level to fall to the
lower mark the viscosity of the liquid can
be calculated. (B) From ground muscle tis-
sue myosin is extracted using high salt so-
lution. Extraction for 20 minutes results in
myosin A, which is characterized by low vis-
cosity, whereas extraction over night results
in myosin B with high viscosity. Addition of
ATP results in decrease of the viscosity of
Myosin B. In protein purifications myosin
B was later shown to consist of actin and
myosin. (C) Myosin solution squirted into
water leads to formation of myosin fibers.
Drying results in myosin fibers structurally
similar to structures seen in electron mi-
crographs of intact muscle [52]. (D) Fibers
are prepared from myosin A and B prepara-
tions. Addition of ATP to actomyosin fibers
results in in vitro contraction. (Illustrations
based on Szent-Gyorgyi, 2004 [51]).

in a changed affinity of all symmetric binding sites for their of the same “cistron”, i.e. same gene, can complement and
corresponding ligand. With this model, several regulatory hence rescue each other. Several models to explain this phe-
phenomena could be explained and enzymes could be catego- nomenon were proposed. The simple model in which com-
rized as “K-systems” where an effector F modifies the affinity plementing mutations at the protein interaction interface en-
of the enzyme for its substrate and “V-systems” where ef- able an otherwise not feasible interaction, was deemed un-
fector binding has no impact on affinity but alters catalytic likely based on the frequency of observed complementation
activity. The model quantitatively explained several observed patterns [67]. Instead, it was proposed that correction acts dis-
effects, including the identical allosteric coefficients of carbon tantly, i.e. not at the interface, but, by mutual correction of the
monoxide and oxygen, despite 250-fold higher absolute affin- overall folding of the involved subunits [67]. A general version
ity of the former for hemoglobin [37]. The authors conclude, of this argument was developed by McGavin in 1968 based on
that “oligomers (are well suited) as mediators of chemical sig- the allosteric model by Monod and colleagues [68]. McGavin
nals [ . . . ] (because) certain desirable physical and physiological considered a multimeric symmetrical protein complex pre-
properties are associated with symmetry, and therefore inaccessible disposed to assume different conformational states. In a ho-
to a monomeric protein” [37]. Protein interactions thus were momeric complex of subunits encoded by a single allele, the
recognized as playing a central role in the dynamic adapta- “damaged” sites are necessarily symmetrical and hence rein-
tion of enzyme activity according to metabolic requirements force each other in stabilizing one or the other conformation.
of the organism [37]. In addition, heterologous interactions In contrast, in a heteromeric complex composed of subunits
of enzymes with nonenzymatic proteins were identified that encoded by different alleles, mutated sites are unsymmetrical
modify catalytic parameters, as demonstrated by the dras- and hence may balance each others effects [68]. McGavin con-
tic increase in Aldolase activity elicited by interaction with cluded that intraallelic complementation should arise with a
F-actin [66]. given frequency, which would also imply evolutionary advan-
Progress in understanding the importance of protein– tages of organizing proteins in multimeric complexes [37,68].
protein interactions led to the idea that perturbation of physi- To this day, the understanding of enzyme regulation by
cal interactions may underlie genetic complementation, in protein–protein and protein–metabolite interactions is a topic
particular interallelic complementation [67]. In interallelic of great interest. Moreover, the recent abundance of system-
complementation different non- or semi-functional alleles atic genetic data on human disease and plant traits relevant


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
1484 P. Braun and A.-C. Gingras Proteomics 2012, 12, 1478–1498

for food production, have increased the urgency for under- the same time the discovery of PKA established the first sig-
standing how genetic variation is mediated on the protein nal transduction cascade.
level to give rise to pathologic or desirable traits (see below). After its discovery, PKA became an intensely studied
protein. In the early 1970s, it was shown that PKA consists
of a tetramer of two catalytic subunits associated to two reg-
ulatory subunits that act as the cAMP sensors [81–83]. Thus,
4.3 The first signal transduction cascade physical interactions were shown to play a critical role in
regulating the activity of signaling proteins. Around the same
During the 1960s, at a time when phosphoproteins were con- time, proteins controlling the specificity of phosphorylase
sidered “biologically inert and ( . . . ) uninteresting” [69], the phosphatase (PP1) via physical interactions were identified;
study of hormone-induced glycogen release led to the discov- studies from Cohen and colleagues identified both protein
ery of the first signal transduction cascade. Work in the 1940s phosphatase inhibitors as well as targeting subunits for PP1,
in Carl and Gerty Coris’ lab had identified glycogen phospho- including a targeting subunit for glycogen and a different
rylase (hereafter phosphorylase), the rate limiting and highly one for myosin [84].
controlled enzyme that catalyzes the first step in the conver- A critical element of signal transduction cascades that re-
sion of glycogen to glucose. They found two interconvertible mained missing for a long time, were the hormone- and
versions of this enzyme; phosphorylase b, which requires ac- growth factor receptors, including those that activated adeny-
tivation by adenosine monophosphate (AMP) and phospho- late cyclase and PKA. Fascinatingly, even though receptors
rylase a, which they assumed had AMP covalently bound as had been studied pharmacologically since the 1920s (see
a prosthetic group. While Coris’ had observed the conversion [85,86]) and detailed theoretical models had been developed to
of phosphorylase a to b [70], and even identified an activity cat- quantitatively describe ligand binding [86], up until the 1970s
alyzing it, their mechanistic explanation of AMP being a pros- the physical existence of receptors was controversial [85, 87].
thetic group, which was removed by the prosthetic group re- A critical step forward was the development of radiolabel-
leasing (PR) enzyme [71], was inconsistent with the absence of ing methods during the 1960s: Isotopic iodine was covalently
nucleosides in crystallized phosphorylase b preparations [69]. attached to receptor agonists and thus enabled quantitative
Fisher and Krebs turned to study phosphorylase in the 1950s determination of ligand binding constants and of receptor
and realized that activation required ATP instead of AMP, molecules on the cell surface. These experiments progressed
which led them to discover that the activated phosphorylase a in parallel for growth factor receptors [88, 89], ion-channel
contained isotopic phosphate when incubated in the presence receptors [90], and G-protein coupled receptors [91]. Radi-
of ␥-32 P-ATP. Phosphorylase thus became the first example olabeled ligands also enabled the purification of receptors,
of an enzyme regulated by phosphorylation. Structural and by tracing their presence during biochemical purification.
functional analysis later revealed that phosphorylase a is pre- Using such an approach, the nicotinic acetylcholine recep-
dominantly a tetramer, whereas the b form occurs mostly as a tor, which is especially abundant in the electric organs of
dimer [72]. Thus, phosphorylase must also be considered the electric ray (Torpedo marmorata) and eel (Electrophorus elec-
first example of a phosphorylation-regulated protein–protein tricus), was first purified [92]. Injection of rabbits with puri-
interaction. Development of an in vitro assay [73] enabled fied receptor resulted in production of antibodies followed by
purification of the upstream kinase (phosphorylase kinase); paralysis of the animals [93, 94]. This result was critical for
the previously described phosphorylase a to b converting en- establishing the functional relevance of the purified agonist-
zyme was identified as the corresponding phosphatase (pro- binding proteins and establishing that receptors and ligands
tein phosphatase 1 or PP1) [74–76]. This early work was fol- were proteins that physically interacted. SDS-PAGE analysis
lowed years later by crystallization studies, paving the way of purified receptors further revealed that they were com-
for the study of phosphorylation-regulated protein–protein posed of five subunits [95], a result that laid the groundwork
interactions (see below). Physiologically, however, important for understanding the structural organization and allosteric
questions remained. Sutherland had shown that hormone regulation of ion channels [87]. In the following years, an in-
stimulation of cells is followed by the production of cyclic tense search and rapid identification of other kinases, phos-
AMP (cAMP) [77, 78], which was critical for the activation phatases, and other downstream effectors of receptors such as
of phosphorylase [77]. While phosphorylase kinase directly heterotromeric G-proteins [96] ensued in many different dis-
regulated phosphorylase, it did not itself bind cAMP, and ciplines. Signal transduction research was born, and protein–
therefore could not be its sensor. This motivated the search protein interactions took a more central role.
for a cAMP-dependent kinase activity, which was identified
by the Krebs laboratory in 1968 [79]. The purified protein
became known as PKA (protein kinase A). As already spec- 5 Interactions in time and space
ulated in the original publication, PKA could phosphorylate
and thereby activate phosphorylase kinase [80]. Thus PKA After the fundamental concepts of molecular biology had
provided a key link between the hormone-induced produc- been laid out, technical innovations in many fields in the
tion of cAMP and the activation of metabolic processes; at late 1960s and early 70s revolutionized the analysis of


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
Proteomics 2012, 12, 1478–1498 1485

molecular biological events. The optimization of cell culture cdc2p protein; antibodies raised against the yeast cdc2p pro-
by virologists [97], development of SDS gel electrophoresis tein were able to recognize a single band in human cells
by biochemists [98], development of monoclonal antibody [116]. This strongly suggested that closely related proteins
technology by immunologists [99–104], refinement of genetic were involved in regulating the cell cycle in evolutionary
screens, and the development of molecular cloning by molec- distant organisms [115]. However, the connection of these
ular biologists [105] provided researchers with a toolbox of kinases to MPF was still unclear. By the end of the 1980s, it
unprecedented capability. Empowered by genetics but chal- became possible to purify MPF, which was found to consist of
lenged by the enormous complexity of human disease, most two proteins. Fascinatingly, the very same antibodies against
notably cancer, simpler systems were adopted for the study yeast cdc2p recognized the frog orthologue of cdc2p as the
of the eukaryotic cell cycle and the molecular causes of cancer first component of MPF [117, 118]. The second protein was
induced by oncogenic viruses. These studies led to major ad- found to be the oscillating cyclin A [119], which activated the
vances in the conceptual appreciation of the central role and kinase in a cell-cycle dependent manner by physically inter-
omnipresence of regulated protein interactions. acting with it [120]. Thus, by the end of the decade, orthogonal
approaches in very different model systems had resulted in
elucidation of the basic concepts of the universal eukaryotic
5.1 Protein interaction controls the eukaryotic cell cell cycle, at the core of which stood the interacting protein
cycle pair of a cyclin and a cyclin-dependent kinase (cdc2) [121].

Cell division was recognized as one of the most defining char-


acteristics of life ever since Virchow concluded that living cells 5.2 Protein interactions’ role in transformation
can only arise from other cells [106]. At the turn of the cen- by oncogenic viruses
tury, embryologists and cytologists had described the macro-
molecular events of cell division using light microscopy, but In parallel to the cell-cycle investigations, the molecular basis
investigation into the molecular underpinnings of this pro- of cellular transformation by oncogenic viruses was intensely
cess started to bear fruit in the 1970s. Exploration of cell cycle studied and led to several important insights about the roles
control was approached in two completely different systems and regulation of protein interactions. By the early 1970s,
in initially separated communities. Geneticists used muta- geneticists had identified viral genes required for transfor-
tion analysis in yeast to identify temperature-sensitive mu- mation by Rous sarcoma virus (RSV), adenovirus, Simian
tants that arrested the cell cycle at specific points [107, 108]. vacuolating virus 40 (SV40), and others. For SV40, a tumor-
Hartwell and colleagues identified a series of cell division specific antigen (large-T) had been discovered in 1963, and
control (CDC) mutants, amid them the budding yeast CDC28 by the late 1970, several associated proteins were known
gene and its fission yeast homologue cdc2. Complementation [64, 122, 123]. The significance of these interactions for cellu-
screening of cdc2 mutants in Paul Nurse’s lab enabled iso- lar transformation was unclear, though. For RSV, attempts to
lation of the coding sequence for cdc2, which was found to raise antibodies against a tumor antigen succeeded in 1976,
encode a protein kinase [109, 110]. and enabled isolation and characterization of the oncogenic
In parallel, biochemists investigated the cell-cycle in v-src protein kinase [124–126]. Identification of a homolo-
oocyte extracts. They observed the breakdown of the nuclear gous kinase encoded in the host genome (c-src) [126], and
envelope, a marker for cell division, when activated cytosolic recognition that the viral protein was a constitutively active
extracts were added to those from inactive oocytes [111]. A version of the proto-oncogenic host kinase, later suggested
critical biochemical activity, named maturation promoting that similar molecular mechanisms underlie etiologically dif-
factor (MPF), was identified as key factor for this process ferent cancers [126]. In 1979 a kinase activity, later shown
[112]. Tim Hunt’s lab later described a protein that appeared to be c-src, was detected in IPs of the SV40 large-T antigen
and disappeared in phase with the cell cycle, and which was [127]. Surprisingly, this kinase preferentially phosphorylated
therefore named cyclin [113]. Even though the first cyclin tyrosine residues. Until then, only phosphorylation on serine
gene was cloned shortly after isolation of the cdc2 cDNA and threonine was documented in eukaryotes. It was there-
[114], by the mid 1980s, the cell cycle studies in oocyte fore unclear whether the identified phosphotyrosine, which
extracts and yeast stood disconnected next to each other, as is recognized today as one of the most significant posttransla-
did cyclin and kinase. tional modifications in growth factor signaling, was the end-
At that time, the widespread of homology between proteins product of the reaction or “merely an intermediate of a reaction
of evolutionary distant organisms was not yet recognized and that could not run to completion in the IP” [127]. Shortly af-
it was assumed that the mechanisms regulating the cell cy- ter, however, tyrosine phosphorylation by epidermal growth
cle in uni and multicellular organisms were fundamentally factor receptor (EGFR) and v-Abl was described [128, 129],
different [115]. This view was shattered when experiments and the initial caution gave way to an intense search for tyro-
in Paul Nurse’s lab identified a human cDNA that comple- sine phosphorylated substrates. With the growing availability
mented the yeast cdc2 mutant. The isolated cDNA not only of sequence information, computational approaches such as
encoded a kinase that exhibited 63% homology to the yeast sequence alignment became increasingly useful [130, 131].


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
1486 P. Braun and A.-C. Gingras Proteomics 2012, 12, 1478–1498

Using a combination of experimental and computational had been observed following a wide array of stimuli [139,140].
analysis, a conserved noncatalytic domain critical for trans- At a time when the cell was still considered an unstructured
formation by src was identified by the Pawson group in 1986 “bag of enzymes” [141] the fact that a single kinase can mediate
[132]. While experimental data indicated the importance of a diversity of signals, some of these causing opposite phys-
the identified domain for transformation, its molecular func- iological effects, was puzzling [141]. More detailed analysis
tion was unclear. However, in 1990 the same group demon- revealed that different external signals can selectively activate
strated that homologous domains (src homology 2, or SH2 a subpopulation of PKA molecules [142]. Subsequent anal-
domains) from several important signaling proteins could ysis of the underlying molecular mechanism revealed that
bind to activated, but not to unstimulated, EGFR [133]. Spec- type II-B PKA was tightly bound to microtubules via a phys-
ulation that the interactions might be regulated by tyrosine ical interaction with a microtubule-associated protein [143].
phosphorylation was later confirmed [134]. This discovery not Genetic experiments confirmed that the physical interactions
only demonstrated that dedicated domains mediate posttrans- with such A-kinase anchoring proteins (AKAPs) are critical
lational modification-dependent interactions, but also sug- for the observed selective activation and the resulting physio-
gested that dynamic regulation of protein interactions was a logical specificity of PKA signals [144]. AKAP-PKA modules
more common phenomenon than was appreciated until then. were the first example of signaling modules that are spatially
In subsequent years, numerous other modification-specific insulated from each other by physical protein–protein interac-
interactions involving structurally diverse protein domains tions [139]. The concept of compartmentalization has become
were identified [135], and the immense combinatorial poten- a major topic in PKA research, and soon encompassed other
tial for signal processing, achieved by concatenating different kinases and signaling molecules, including raf-ras-MAPK sig-
interaction domains, was recognized. naling [145]. This remains an active area of research and more
By the late 1980s, work on the src oncogenes had iden- compartmentalized signaling structures are likely to be iden-
tified growth promoting host pathways and suggested that tified in coming years [141].
retroviruses cause cellular transformation by activating these Thus, in the 1970s and 1980s, a multitude of technical
pathways. Despite the presence of src activity in SV40 large-T advances and the confluence of the hitherto largely sepa-
IPs though, the mechanism by which DNA tumor viruses rated fields of genetics, biochemistry, and cell biology brought
such as SV40 and adenovirus caused transformation re- about a dramatic boost in the understanding of molecular cel-
mained largely elusive. Several physical interaction partners lular processes. In the course of these experiments, physical
for adenovirus E1A and large-T antigen, identified as bands protein interactions appeared to play key roles in all investi-
on SDS-PAGE gels of IPs had been characterized and addi- gated processes, including the molecular events underlying
tional proteins remained to be studied. One of the unknown one of the most deadly human diseases. These advances in
proteins found in IP of the adenoviruses E1A oncoprotein the conceptual understanding of the importance of protein
had a molecular weight of approximately 105 kD. Around the interactions were brought about by numerous technological
same time, the Weinberg laboratory had identified the first innovations that facilitated interaction analysis and later acted
example of a new class of cancer genes–the retinoblastoma synergistically with the completion of genome sequences for
tumor suppressor gene (Rb). The molecular function of the several organisms.
encoded 105 kD retinoblastoma protein (pRb), however, was
unclear. Reciprocal IPs using antibodies against pRb or the
E1A-associated protein revealed the identity of both proteins, 6 Genome ready technologies
resulting in the first characterization of a physical interaction
between a viral oncogene and a human tumor suppressor At the beginning of the 1990s, protein interactions were
gene in 1988 [136]. This suggested that both positive and neg- widely recognized as critical for most cellular processes, and
ative growth regulators do not necessarily function through additional technological innovations facilitated their analysis.
separate pathways but “may directly confront one another This resulted in an explosion of papers describing interacting
through the interaction of their encoded proteins” [136]. In subse- protein pairs (Fig. 5). An important innovation, which set the
quent years, additional examples of this “direct confrontation” foundation for essentially all currently used high-throughput
were discovered, including interactions of viral oncogenes (HT) interaction assays, was the application of gene fusions
with pRb and p53, which in 1989 was recognized as a tumor to the study of proteins. The first gene fusions had been used
suppressor protein [137], and interactions between pro and to study gene regulation [146]. In the 1980s, the technology
antiapoptotic members of the BCL-2 protein family [138]. was applied to proteins: a cDNA for the protein of interest was
genetically attached to a second cDNA encoding a protein that
5.3 Spatial control of interactions in signaling provides additional functionality. The “hybrid protein” solved
modules a nagging problem at the time when genes involved in diverse
processes were being identified, but no biochemical assay or
In addition to the temporal regulation of signaling, the im- antibody existed for the encoded proteins, which largely pre-
portance of spatial regulation became apparent during the vented their purification and biochemical characterization. In
continued investigation of PKA. By the 1980s, PKA activation 1979, the first hybrid protein was described, consisting of a


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
Proteomics 2012, 12, 1478–1498 1487

the tip of a fine capillary directly into the mass spectrometer


[157]. ESI affords several advantages over MALDI, not least
the ability to directly couple the separation of peptide mixture
by high-performance liquid chromatography to their mass
spectrometric identification. The invention of tandem MS en-
abled fragmentation and hence sequencing of single peptides
[158, 159]. However, largely due to the difficulties in de novo
protein sequencing by MS, Edman degradation remained a
critical technique for the de novo identification of unknown
proteins until the beginning of the genomic revolution.
The identification of novel interacting proteins thus was
still challenging. To complement approaches based on the
biochemical enrichment of proteins, a number of alternative
approaches were developed. In 1989, Fields and Song pub-
lished a method that circumvented biochemistry altogether,
and thus greatly facilitated the identification of novel protein
interaction partners: the yeast-2-hybrid (Y2H) system [160].
Building on studies that demonstrated that the DNA bind-
ing and activating functions of the yeast transcription factor
GAL4 can be separated, they developed a system in which a
Figure 5. Number of publications published each year on the
physical interaction of two proteins was detected genetically
topic "protein-protein interaction". (Source: Science citation in-
by activation of a selectable reporter in yeast. This straight-
dex)
forward system could be used to screen large cDNA libraries
using a simple growth-selection based assay, and novel inter-
bacterial protein of unknown function and β-Galactosidase, actors could be identified using routine DNA sequencing. In
which could be used to detect and purify the hybrid protein subsequent years, many key interactions were identified by
using a routine colorimetric assay [147]. The genetic fusion Y2H, including one between the mammalian oncogene ras
with a partner exhibiting high affinity for a ligand provided the and its downstream kinase raf [161]. A foreshadowing of its
additional advantage that an affinity step could be included in future importance came in 1996, when the Y2H system was
the purification procedure [148]. The power of this approach used to construct the first proteome-scale “protein linkage
was quickly realized, and many dedicated affinity-fusion tags map”: that of bacteriophage lambda containing 25 interac-
were developed, including glutathione-S-transferase (GST), tions [162]. Importantly, the development of Y2H was also
which simplified elution with an economical, dialyzable elu- a proof of concept, which encouraged the development of a
ent [149]. Combination of fusion tags, were also introduced, large number of technologies for protein interaction analysis
which enabled the generation of proteins of higher purity that are based on the reconstitution of a functional protein by
and led to the development of tandem affinity purification protein–protein interactions [163–167]. Several of these are
for proteome-scale protein complex analysis [150–153]. Fu- described in chapters by Lam and Stagljar, Braun, and Taylor
sion proteins with peptide sequences recognized by specific and Wrana in this issue.
monoclonal antibodies, e.g. the Hemaglutinin (HA) peptide During the 1990s, methods based on affinity purification
of the influenza virus (HA-tag) [154], enabled both detection of interacting proteins, on genetic approaches such as the
and purification of the protein using immunological proce- Y2H system and other interaction detection strategies were
dures, and hence greatly facilitated the characterization of employed by many different labs to address questions in es-
proteins and the identification of their interactions. sentially all fields of molecular biological inquiry. The true
The value of protein purification for discovery of new in- power of all these approaches, however, needed additional
teraction partners rests in the ability to identify the unknown ingredients to blossom into the power-house technologies
partner. This can be done using techniques such as Edman that they are today.
degradation [155] and more recently mass spectrometry (MS).
A limitation of early mass spectrometers was that large macro-
molecules such as peptides or proteins are difficult to ionize. 7 The genomic revolution–protein
Toward the end of 1980s, two new ionization methods for interaction networks
biological samples were introduced, which transformed the
field of biological MS and which are still in use today. Matrix- Against the backdrop of heavy criticism for being too ex-
assisted laser desorption ionization (MALDI) is based on ion- pensive, “bad big biology”, “technically impossible” and “of
ization of molecules embedded in a crystallized matrix by UV limited scientific value” [168], the Human Genome Project
light laser [156]. In electrospray ionization (ESI), a high volt- (HGP) was officially launched in 1990. The resulting ge-
age is applied to an analyte containing solvent exiting through nomics revolution also majorly affected the analysis of


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
1488 P. Braun and A.-C. Gingras Proteomics 2012, 12, 1478–1498

protein–protein interactions. A key development was the regulation of mRNA expression or presence in genomes, are
identification of (nearly) all genes encoded in a genome. correlated for physically interacting proteins [206]. Many of
Genomic and expressed sequence tag (EST) sequences pro- the discovered correlations, as well as structural information,
vided a template for the construction of genome-scale cDNA have later been used to predict physical interactions of pro-
collections (e.g. the mammalian gene collection, MGC) and teins using bioinformatic approaches [207–209].
open reading frame (ORF) collections [169–176]. In contrast The exclusive focus of the first reports describing large-
to cDNA libraries, in ORF collections only the protein cod- scale approaches to protein–protein interaction analysis was
ing sequence of a transcript is encoded by a plasmid and development of hypotheses for unknown proteins [180]. Soon
not the 5 and 3 untranslated regions of the cDNA. There- after publication of the first data sets, however, the network
fore, the encoded proteins can be directly fused to tags and structure and global interconnectivity of all proteins was no-
hence can be used in a variety of functional experiments. In ticed and the network itself was investigated [210]. Analysis
addition, during construction of indexed cDNA or ORF col- of protein–protein interaction networks using computational
lections, biases and differential representation of clones are and statistical tools has since then contributed important in-
largely eliminated, such that a given sequence is present only sights about the large-scale organization of biological systems
once in the collection. This elimination of redundancy has en- and how networks mediate genotype to phenotype relation-
abled complex experiments, including the analysis of protein ships (see also [211]). Structurally, binary interactome maps,
interactions, which would otherwise have been prohibitively i.e. maps based on binary assays like Y2H that detect pre-
expensive. The development of recombinational cloning fur- dominantly direct interactions, were found to have a so-called
ther facilitated the HT transfer of ORFs into expression vec- scale-free topology [212]. In networks of this topology, most
tors for functional experiments, including global Y2H screens proteins (nodes) have one or few interactions (edges) and few
[169, 177, 178]. For MS, the availability of complete gene cat- proteins, so called hubs, have a very large number of inter-
alogues provided an additional massive advantage: instead action partners. Networks with this topology, which is also
of sequencing proteins de novo using peptide fragmentation found for other complex networks such as the internet or so-
spectra, which to this day is not automated, the fragmentation cial networks, are resilient against failure of random nodes,
spectra could be matched against the translation products of e.g. by mutation, but sensitive to targeted attack of the hubs
all genes using easily automatable algorithms [159]. [213]. Fascinatingly, in both plants and human, proteins of
Interactome analysis thus benefited immensely from the viral, bacterial, and fungal pathogens were all found to target
HGP. At the same time, protein interactions offer an excit- such hub proteins [214–218]. In addition to providing insights
ing and efficient route to assign biological function to the into the structural organization of interactome networks, the
thousands of unknown proteins that were uncovered in ev- relationship between physical and genetic interactions in bi-
ery sequenced genome [179, 180]. With this, the focus of in- ological systems is becoming increasingly clear. In yeast, a
quiry shifted from “which proteins are involved in my biolog- positive correlation between the number of direct physical
ical phenomenon?” to “which processes are all the encoded interaction partners of a protein and the number of pheno-
proteins involved in?” [69], and large-scale protein interac- types of the corresponding genetic null was observed [219],
tion analysis became an important tool for answering this suggesting that at least hubs in binary networks participate
question. in diverse processes. For human diseases, proteins involved
in a common disease were found more to likely interact than
expected by chance [220]. Conversely, for several diseases, the
8 Interactome networks protein products of disease causing genes were found to phys-
ically interact and form highly connected disease “modules”
Confronted with an immense number of genes encoding [221, 222]. Further, network analysis has been successfully
proteins of unknown function, and equipped with automat- used in defining novel cancer and disease genes [223, 224].
able protein interaction assays, global interaction mapping Whereas Y2H-based interactome maps revealed the scale-free
experiments started while sequencing of the human genome topology of the network of direct interactions, AP-MS-based
was still underway. The first systematic protein–protein in- interactome analysis illuminated the modularity of the in-
teraction maps were published in 2000 for Y2H [181–184] teractome and revealed molecular machines [187, 188, 225].
and in 2002 for AP-MS analysis [185–188], followed soon by Combining protein complex information with data on ge-
large-scale interaction maps using different technologies for netic interactions in yeast revealed that proteins within com-
C. elegans, D. melanogaster, H. sapiens, several microorgan- plexes are characterized by similar patterns of genetic interac-
isms, and recently higher plants [189–198]. Especially for S. tions [226–230]. In recent years, protein–protein interaction
cerevisiae, which served as test case for essentially all large- network information has played an increasing role in the
scale approaches, other functional data were collected in a analysis of genetic disease data, e.g. from genome-wide asso-
systematic fashion including localization [199], transcription ciation studies [231, 232]. Moreover, computational integra-
[200], phenotypes [201, 202], genetic interactions [203], and tion of interaction information with transcriptional informa-
kinase-substrate interactions [204, 205]. These analyses have tion from tumors was shown to have predictive value for the
led to the realization that other biological properties, e.g. the classification of tumor samples ([233] and Taylor et al., this


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
Proteomics 2012, 12, 1478–1498 1489

issue) Thus, interactome network analysis has become an quality are critical. Due to the nature of MS data, very dif-
important tool for understanding the organization of cellular ferent solutions from those for binary assays are being devel-
systems [211, 234, 235]. oped. Quantitative information regarding the abundance of
In addition to their biological interpretation, data quality each prey is embedded in every MS experiment. The use of
became an important topic in the field of protein interaction this quantitative (label-free) information can be harnessed to
analysis. While initially it was argued that interaction data provide confidence score for each bait-prey interaction pair,
from high-throughput approaches are inferior to those ob- by performing modeling either against a set of well-defined
tained in small-scale studies, today it has been shown that negative control experiments, or across large sets of AP-MS
high-throughput approaches can provide data of a quality samples [225, 242–246]. This type of approach is pursued by
identical to those supported by multiple small-scale studies several labs and covered in greater detail in the reviews by
collected in manually curated public databases [193,219,236]. Nesvizhskii, and Dunham et al., in this issue. Alternatively,
This is partly due to considerable progress in the concep- approaches exploiting isotopic labeling as a means to iden-
tual understanding of the data produced by different tech- tify true interactors have also led to significant breakthroughs
nologies, and aided by the development of quantitative ex- (e.g. [247–250], and reviewed by Trinkle-Mulcahy et al., this
perimental approaches to evaluate interactome data sets. For issue).
binary assays, the use of standardized reference sets by dif- While most HT AP-MS efforts have thus far focused on the
ferent laboratories to benchmark the assays has started to use of epitope tags as affinity handles, an alternative approach
increase transparency and comparability of data [236–238]. is to employ antibodies specific for the endogenous proteins
Moreover, development of a quantitative framework for the themselves. While a number of difficulties (e.g. procurement
evaluation of interactome maps, also based on reference sets, of antibodies to each target and design of appropriate con-
led to the insight that currently available systematic binary trols) are associated with this approach (reviewed in Dunham
network maps only cover a small fraction of the respec- et al., this issue), the growing availability of both natural and
tive interactomes of an organism [193, 219, 236, 237, 239]. synthetic antibodies as affinity reagents [251–254] enables
By taking into account the limitations of interactome map- researchers to analyze protein interactions in more physio-
ping experiments, namely the completeness of the ORF col- logical conditions. Recently, a study using greater than 1800
lection, the assay sensitivity, and the degree of saturation of antibodies against transcription coregulator proteins demon-
the screen, it is possible to determine what fraction of a strated that this is a viable approach for high-throughput
given interactome has been mapped. By this method, the interaction mapping, provided that stringent data filtering
binary interactomes of yeast, human, and plant have been is applied [255].
estimated to consist of approximately 18 000, 160 000, and
320 000 protein–protein interactions, respectively, of which
between 2% and 20% have been mapped by systematic high- 9 Protein–protein interactions
quality maps [193, 219, 236]. Alternative approaches to es- and therapeutics
timate the interactome size obtain slightly higher, although
overall similar estimates [240,241]. Thus, increasing the over- Because of the ubiquity of protein–protein interactions and
all sensitivity of interactome maps without sacrificing data the diversity of mediated biological functions, they consti-
quality is a pressing challenge in the field as discussed tute attractive targets for the modulation of biological sys-
in more detail by Braun in this issue [167]. For S. cere- tems. Therapeutic antibodies against secreted proteins and
visiae, Y2H-based binary interactome maps were comple- receptors are an important and rapidly growing segment of
mented by systematic data sets acquired with other tech- the prescription drug market [256, 257]. Moreover, intracel-
nologies including split-ubiquitin and protein complemen- lular modulators of protein–protein interactions, e.g. tubulin
tation assay [196, 197]. The development of standardized in- binders such as colchicine or taxol, have been used clini-
teraction assay benchmarking and strategies for systematic cally for many years [258,259]. Nonetheless, developing small
confidence scoring of individual interactions are important molecule modulators of intracellular regulatory interactions
developments that facilitate the comparison and integration is challenging [260]. In contrast to the deep substrate bind-
of interactome maps acquired with different technologies ing pockets of enzymes, interaction surfaces are usually more
[167, 237]. flat and extended. Nonetheless, important progress and proof
Interactome analysis using AP-MS is dealing with simi- of concept experiments have demonstrated the general fea-
lar challenges regarding the need to increase assay sensitiv- sibility of targeting protein–protein interactions with small
ity, to standardize and compare results produced in different molecules. The general goal is to identify so-called hot spots
laboratories, and develop methods for robust and compara- on the surface that either contribute a large portion of the total
ble determination of interaction data quality. Single-tag pu- binding energy, or are amenable to steric interference with a
rifications are increasingly used to capture more transient small molecule compound [261]. While identification of such
interactions and the detection sensitivity of instruments is hot spots is often done by scanning mutagenesis, genetic
improving spectacularly. While this is undoubtedly exciting, approaches such as the reverse Y2H, in which interaction-
widely accepted and standardized methods to assess data deficient full-length alleles can be isolated by selection in


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
1490 P. Braun and A.-C. Gingras Proteomics 2012, 12, 1478–1498

yeast, could provide a cost-effective alternative [262,263]. Even has been shown for both human disease and plants, there is
after identification of hot spots, finding a small molecule in- a great potential for even simple variation of primary protein
hibitor by rational design or HT screening remains difficult. sequence, e.g. due to natural genetic variation or pathologic
A large part of the challenge is that compounds in exist- or experimental selection, to modulate interactions qualita-
ing libraries are structurally biased towards classical targets tively (all or none) or quantitatively (interaction strength) and
for which they were initially developed, such as enzymes, thereby affect phenotypes [263, 269–271]. Thus knowledge
ion channels and G-protein coupled receptors. Nonetheless, of interaction wiring not only feeds global network analy-
small molecule inhibitors for several protein–protein inter- sis, but can provide the foundation for an understanding of
actions have been identified, e.g. disrupting the interaction the specific molecular mechanisms underlying pathologic or
between human p53 with human MDM2 or between the hu- desirable phenotypes. However, at the same time, interac-
man papillomavirus (HPV) E1 and E2 proteins [260,264]. The tion maps need to become more comprehensive to optimally
available examples indicate that inhibitors of protein interac- complement modern genomic data, and interaction analysis
tions, indeed, do not exhibit structural similarity with other technologies need to become more versatile to probe protein–
classes of small molecule inhibitors [260]. Thus, it is likely protein interactions under different physiological conditions.
that protein–protein interaction “inhibitors are likely more iso- In order to overcome current limitations, new technologies
lated in chemical space” [260]. In addition to expanding ex- for protein–protein interaction analysis will be required. The
isting chemical libraries, availability of reliable information articles in this issue of PROTEOMICS provide an overview
on protein interactions and their role in physiological and over state-of-the-art technologies and ongoing technology de-
pathophysiological processes is critical for the development velopment for analysis of protein–protein interactions and
protein–protein interaction-directed therapeutics. their dynamic regulation, and for strategies toward obtaining
high-quality interactome maps of increasing completeness.

10 Summary and outlook The authors have declared no conflict of interest.

In the 20th century, the basic principles of molecular bio-


logical regulation have been discovered and described. Dur- 11 References
ing this time, physical interactions among proteins were
[1] Durbin, R. M., Abecasis, G. R., Altshuler, D. L., Auton, A.
found to mediate a vast range of regulatory functions in es-
et al., A map of human genome variation from population-
sentially every biological process. In the first decade of the scale sequencing. Nature 2010, 467, 1061–1073.
21st century, insights into the global structure of protein–
[2] Vickery, H. B., Origin of the word protein. Nature 1951, 168,
protein interaction networks were gained, and the relation-
244.
ship between genetics and the physical interactions of the
[3] Keith, L. M., Louis Pasteur (1822–1895) — chance and the
encoded genes is emerging. However, big challenges re-
prepared mind. Trends Biotechnol. 1995, 13, 511–515.
main. As we have discussed, the biochemical relationships
between proteins and the functional consequences of phys- [4] Hein, G. E., The Liebig-Pasteur controversy: vitality without
vitalism. J. Chem. Edu. 1961, 38, 614.
ical interactions are profoundly more complicated than our
current understanding for the vast majority of proteins. Sim- [5] Payen and Persoz, Mémoire sur la diastase, et les prin-
ilarly our bioinformatic tools—ranging from databases to cipaux produits de ses réactions. Ann. de Chimie et de
Physique 1933, 53, 73–92.
quantitative dynamic process models—need to become more
suited to represent and productively analyze this biochemical [6] Florkin, M., Discovery of pepsin by Theodor Schwann. Re-
complexity, once it is experimentally accessible on a larger vue Medicale De Liege 1957, 12, 139–144.
scale. [7] Kühne, W., Ueber das Trypsin. Verhandlungen des
Network analysis is well suited to deal with complexity naturhistorisch-medicinischen Vereins zu Heidelberg 1876,
and aims to detect large-scale organizational principles of 1, 194–198.
complex systems by intentionally ignoring, e.g. mechanis- [8] Hedin, S. G., Trypsin and antitrypsin. Biochem. J. 1906, 1,
tic biochemical, details [265, 266]. Even though it is a young 474–483.
field, as briefly discussed above, network analysis has already [9] Northrop, J. H., The inactivation of trypsin : II. the equilib-
contributed important insights to the understanding of bi- rium between trypsin and the inhibiting substance formed
ological systems (see also [211, 267]). A central question of by its action on proteins. J. Gen. Physiol. 1922, 4, 245–260.
biological research, especially in our modern era, is how ge- [10] Sumner, J. B., The isolation and crystallization of the en-
netic variation leads to desirable and pathologic phenotypes. zyme urease. Preliminary paper. J. Biol. Chem. 1926, 69,
Genetics, although its tools have become more powerful than 435–441.
ever before in the history of biology, can only provide part of [11] Bradberry, S., Ricin and abrin. 2007, 35, 576–577.
the answer. Genetic information needs to be complemented [12] Ehrlich, P., Experimentelle Untersuchungen über Immu-
with wiring diagrams and maps of how the encoded proteins nität. II. Ueber Abrin. Deutsche Medizinische Wochenschrift
functionally and physically interact with each other [268]. As 1891, 17, 1218–1219.


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
Proteomics 2012, 12, 1478–1498 1491

[13] Umemura, R., The relation between the absorption of anti- from enzymic hydrolysates. Biochem. J. 1953, 53, 366–
bodies and the isolated protein bodies. J. Immunol. 1921, 374.
6, 205–222. [34] Hauptman, H. A., History of x-ray crystallography.
[14] Buchner, E., Alkoholische Gärung ohne Hefezellen Chemometr. Intell. Lab. Sys. 1991, 10, 13–18.
(Vorläufige Mitteilung). Berichte der Deutschen Chemis- [35] Kendrew, J. C., Structure and function in myoglobin and
chen. Gesellschaft 1897, 30, 117–124. other proteins. Fed. Proc. 1959, 18, 740–751.
[15] Kastle, J. H., On the vital activity of the enzymes. Science [36] Perutz, M. F., Structure of hemoglobin. Brookhaven Symp.
1901, 13, 765–771. Biol. 1960, 13, 165–183.
[16] Loeb, J., The proteins and colloid chemistry. Science 1920, [37] Monod, J., Wyman, J., Changeux, J. P., On the nature of
52, 449–456. allosteric transitions: a plausible model. J. Mol. Biol. 1965,
[17] Loew, O., On the chemical nature of enzymes. Science 12, 88–118.
(New York, N.Y.) 1899, 10, 955–961. [38] Perutz, M. F., Fersht, A. R., Simon, S. R., Roberts, G. C.
[18] Miller, E. R., The chemical nature of enzymes. Science K., Influence of globin structure on state of heme .2. Al-
1929, 70, 356–357. losteric transitions in methemoglobin. Biochemistry 1974,
[19] Svedberg, T., Chirnoaga. E., The molecular weight 13, 2174–2186.
of hemocyanin. J. Am. Chem. Soc. 1928, 50, 1399– [39] Watson, J. D., Crick, F. H. C., Molecular structure of nucleic
1411. acids - a structure for deoxyribose nucleic acid. Nature
[20] Svedberg, T., Mass and size of protein molecules. Nature 1953, 171, 737–738.
1929, 123, 871. [40] Aristotle (Peck, A. L., Foster, E. S., translators), Parts of
[21] Inga, B., Eriksson, Q., Svedberg, T., The molecular weights Animals, Movement of Animals, Progression of Animals.,
and pH-stability regions of the hemocyanins. Biol. Bull. Loeb Classical Library No. 323, 1937.
1936, 71, 498–547. [41] Norris, R., Researches on muscular irritability and the re-
[22] van Holde, K. E., Reflections on a century of protein chem- lations which exist between muscle, nerve and blood.
istry. Biophys. Chem. 2003, 100, 71–79. J. Anat. Physiol. 1867, 1, 217–236.

[23] Bergmann, M., The structure of proteins in relation to bio- [42] von Muralt, L., Edsall, J. T., Double refraction of myosin and
logical problems. Chem. Rev. 1938, 22, 423–435. its relation to the structure of the muscle fibre. Transac.
Faraday Soc. 1930, 26, 0837–0863.
[24] Sanger, F., Tuppy, H., The amino-acid sequence in the
phenylalanyl chain of insulin .1. The identification of lower [43] Engelhardt, W. A., Ljubimowa, M. N., Myosine and
peptides from partial hydrolysates. Biochem. J. 1951, 49, adenosinetriphosphatase. Nature 1939, 144, 668–669.
463–481. [44] Bailey, K., Myosin and adenosine triphosphatase.
[25] Sanger, F., Tuppy, H., The amino-acid sequence in the Biochem. J. 1942, 36, 121–139.
phenylalanyl chain of insulin .2. The investigation of pep- [45] Needham, D. M., The adenosine triphosphatase activity of
tides from enzymic hydrolysates. Biochem. J. 1951, 49, myosin preparations. Biochem. J. 1942, 36, 113–120.
481–490. [46] Dainty, M., Kleinzeller, A., Lawrence, A. S. C., Miall, M.
[26] Sanger, F., Some peptides from insulin. Nature 1948, 162, et al., Studies on the anomalous viscosity and flow-
491–492. birefringence of protein solutions iii. Changes in these
[27] Sanger, F., Some chemical investigations on the structure properties of myosin solutions in relation to adenosine
of insulin. Cold Spring Harb. Symp. Quant. Biol. 1950, 14, triphosphatase and muscular contraction. J. Gen. Physiol.
153–160. 1944, 27, 355–399.

[28] Tiselius, A., A new apparatus for electrophoretic analysis [47] Banga, I., Szent-Györgyi A. G., Preparation and properties
of colloidal mixtures. Trans. Faraday Soc. 1937, 33, 0524– of myosin A and B. Stud. Inst. Med. Chem. Univ. Szeged.
0530. 1942, I:5–15.

[29] Tiselius, A., Adsorption analysis of amino acid mixtures. [48] Guba, F., Observations on myosin and actomyosin. Stud.
Adv. Protein Chem. 1947, 3, 67–93. Inst. Med. Chem. Univ. Szeged. 1943, III:40–45.

[30] Consden, R., Gordon, A. H., Martin, A. J. P., Qualitative [49] Jakus, M. A., Hall, C. E., Studies of actin and myosin.
analysis of proteins: a partition chromatographic method J. Biol. Chem. 1947, 167, 705–714.
using paper. Biochem. J. 1944, 38, 224–232. [50] Kesztyus, L., Nikodemusz, S., Szilagyi, T., Antigenic activity
[31] Sanger, F., The arrangement of amino acids in proteins. of myosin and actin. Nature 1949, 163, 136.
Adv. Protein Chem. 1952, 7, 1–67. [51] Szent-Gyorgyi, A. G., Milestone in physiology - the early
[32] Sanger, F., Thompson, E. O. P., The amino-acid sequence history of the biochemistry of muscle contraction. J. Gen.
in the glycyl chain of insulin .1. The identification of lower Physiol. 2004, 123, 631–641.
peptides from partial hydrolysates. Biochem. J. 1953, 53, [52] Straub, F. B., Actin. Stud. Inst. Med. Chem. Univ. Szeged
353–366. 1942, II:3–15.
[33] Sanger, F., Thompson, E. O. P., The amino-acid sequence in [53] Straub, F. B., Actin, II. Stud. Inst. Med. Chem. Univ. Szeged
the glycyl chain of insulin .2. The investigation of peptides 1943, III:23–37.


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
1492 P. Braun and A.-C. Gingras Proteomics 2012, 12, 1478–1498

[54] Straub, F. B., Note on the work of f. Bruno Straub con- verting enzyme of rabbit skeletal muscle. Biochim. Bio-
cerning ‘adenosine triphosphate. The functional group of phys. Acta 1956, 20, 150–157.
actin’. Biochim. Biophys. Acta 1989, 1000, 179. [75] Graves, D. J., Fischer, E. H., Krebs, E. G., Specificity stud-
[55] Straub, F. B., Feuer, G., [adenosine triphosphate, the func- ies on muscle phosphorylase phosphatase. J. Biol. Chem.
tional group of actin]. Kiserl Orvostud 1950, 2, 141–151. 1960, 235, 805–809.
[56] Straub, F. B., Feuer, G., Adenosinetriphosphate. The func- [76] Brandt, H., Capulong, Z. L., Lee, E. C., Purification and prop-
tional group of actin. 1950. Biochim. Biophys. Acta 1989, erties of rabbit liver phosphorylase phosphatase. J. Biol.
1000, 180–195. Chem. 1975, 250, 8038–8044.
[57] Huxley, H. E., The mechanism of muscular contraction. [77] Davoren, P. R., Sutherland, E. W., The effect of L-
Clin. Orthop. Related Res. 2002, S6–S17. epinephrine and other agents on the synthesis and re-
[58] Huxley, H. E., Mechanism of muscular contraction. Science lease of adenosine 3’,5’-phosphate by whole pigeon ery-
1969, 164, 1356–1366. throcytes. J. Biol. Chem. 1963, 238, 3009–3015.

[59] Fletcher, D. A., Mullins, D., Cell mechanics and the cy- [78] Rall, T. W., Sutherland, E. W., Formation of a cyclic adenine
toskeleton. Nature 2010, 463, 485–492. ribonucleotide by tissue particles. J. Biol. Chem. 1958, 232,
1065–1076.
[60] Beadle, G. W., Tatum, E. L., Genetic control of biochemical
reactions in neurospora. Proc. Natl. Acad. Sci. USA 1941, [79] Walsh, D. A., Perkins, J. P., Krebs, E. G., An adeno-
27, 499–506. sine 3’,5’-monophosphate-dependant protein kinase from
rabbit skeletal muscle. J. Biol. Chem. 1968, 243, 3763–
[61] Horowitz, N. H., Beadle, George, Wells (1903-1989) - obit-
3765.
uary. Genetics 1990, 124, 1–6.
[80] Sutherland, E. W., Robison, G. A., Role of cyclic AMP in
[62] Sanadi, D. R., Littlefield, J. W., Studies on alpha-
control of carbohydrate metabolism. Diabetes 1969, 18,
ketoglutaric oxidase .2. Purification and properties. J. Biol.
797–819.
Chem. 1952, 197, 851–862.
[81] Taylor, S. S., cAMP-dependent protein kinase. Model
[63] Koike, M., Reed, L. J., Carroll, W. R., Alpha-keto acid dehy-
for an enzyme family. J. Biol. Chem. 1989, 264, 8443–
drogenation complexes .1. Purification and properties of
8446.
pyruvate and alpha-ketoglutarate dehydrogenation com-
plexes of Escherichia coli. J. Biol. Chem. 1960, 235, 1924– [82] Corbin, J. D., Keely, S. L., Soderling, T. R., Park, C. R.,
1930. Hormonal regulation of adenosine 3’,5’-monophosphate-
dependent protein kinase. Adv. Cyclic Nucl. Res. 1975, 5,
[64] Hayakawa, T., Muta, H., Hirashima, M., Ide, S. et al., Isola-
265–279.
tion and properties of pyruvate and ␣-ketoglutarate dehy-
drogenation complexes from pig heart muscle. Biochem. [83] Beavo, J. A., Bechtel, P. J., Krebs, E. G., Mechanisms of
Biophys. Res. Commun. 1964, 17, 51–56. control for cAMP-dependent protein kinase from skeletal
muscle. Adv. Cyclic Nucl. Res. 1975, 5, 241–251.
[65] Frieden, C., Protein-protein interaction and enzymatic ac-
tivity. Annu. Rev. Biochem. 1971, 40, 653–696. [84] Hubbard, M. J., Cohen, P., On target with a new mecha-
nism for the regulation of protein-phosphorylation. Trends
[66] Arnold, H., Pette, D., Binding of aldolase and triosephos-
Biochem. Sci. 1993, 18, 172–177.
phate dehydrogenase to F-actin and modification of cat-
alytic properties of aldolase. Eur. J. Biochem. 1970, 15, [85] Lefkowitz, R. J., Historical review: a brief history and per-
360–366. sonal retrospective of seven-transmembrane receptors.
Trends Pharmacol. Sci. 2004, 25, 413–422.
[67] Crick, F. H. C., Orgel, L. E., The theory of inter-allelic com-
plementation. J. Mol. Biol. 1964, 8, 161–165. [86] Rang, H. P., The receptor concept: pharmacology’s big idea.
[68] McGavin, S., Interallelic complementation and allostery. J. Brit. J. Pharmacol. 2006, 147, S9–S16.
Mol. Biol. 1968, 37, 239–242. [87] Changeux, J. P., Allosteric receptors: from electric organ to
[69] Fischer, E. H., Phosphorylase and the origin of reversible cognition. Annual Review of Pharmacology and Toxicol-
protein phosphorylation. Biol. Chem. 2010, 391, 131–137. ogy, Annual Reviews, Palo Alto 2010, pp. 1–38.

[70] Cori, G. T., Cori, C. F., The enzymatic conversion of [88] Cuatreca, P., Insulin-receptor interactions in adipose tissue
phosphorylase-a to phosphorylase-b. J. Biol. Chem. 1945, cells: direct measurement and properties. Proc. Natl. Acad.
158, 321–332. Sci. USA 1971, 68, 1264–1268.

[71] Cori, G. T., Green, A. A., Crystalline muscle phosphorylase [89] Carpenter, G., Cohen, S., I125 labeled human epidermal
ii. Prosthetic group. J. Biol. Chem. 1943, 151, 31–38. growth-factor - binding, internalization, and degradation
in human fibroblasts. J. Cell Biol. 1976, 71, 159–171.
[72] Barford, D., Hu, S. H., Johnson, L. N., Structural mecha-
nism for glycogen-phosphorylase control by phosphoryla- [90] Fambroug, D. M., Hartzell, H. C., Acetylcholine receptors -
tion and AMP. J. Mol. Biol. 1991, 218, 233–260. number and distribution at neuromuscular junctions in rat
diaphragm. Science 1972, 176, 189–191.
[73] Fischer, E. H., Krebs, E. G., Conversion of phosphorylase-b
to phosphorylase-a in muscle extracts. J. Biol. Chem. 1955, [91] Lefkowitz, R. J., Roth, J., Pricer, W., Pastan, I., ACTH recep-
216, 121–132. tors in the adrenal: specific binding of ACTH-125i and its
relation to adenyl cyclase. Proc. Natl. Acad. Sci. 1970, 65,
[74] Krebs, E. G., Fischer, E. H., The phosphorylase b to a con-
745–752.


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
Proteomics 2012, 12, 1478–1498 1493

[92] Miledi, R., Molinoff, P., Potter, L. T., Isolation of cholinergic [111] Masui, Y., Markert, C. L., Cytoplasmic control of nuclear
receptor protein of torpedo electric tissue. Nature 1971, behavior during meiotic maturation of frog oocytes. J. Exp.
229, 554–557. Zool. 1971, 177, 129-&.
[93] Patrick, J., Lindstro, J., Autoimmune response to acetyl- [112] Wasserman, W. J., Masui, Y., Cytoplasmic factor promoting
choline receptor. Science 1973, 180, 871–872. oocyte maturation - its extraction and preliminary charac-
[94] Sugiyama, H., Benda, P., Meunier, J. C., Changeux, J. P., Im- terization. Science 1976, 191, 1266–1268.
munological characterization of cholinergic receptor pro- [113] Evans, T., Rosenthal, E. T., Youngblom, J., Distel, D. et al.,
tein from electrophorus-electricus. FEBS Lett. 1973, 35, Cyclin - a protein specified by maternal messenger-RNA in
124–128. sea-urchin eggs that is destroyed at each cleavage division.
[95] Hucho, F., Changeux, J.-P., Molecular weight and quater- Cell 1983, 33, 389–396.
nary structure of the cholinergic receptor protein extracted [114] Pines, J., Hunt, T., Molecular-cloning and characterization
by detergents from electrophorus electricus electric tissue. of the messenger-RNA for cyclin from sea-urchin eggs.
FEBS Lett. 1973, 38, 11–15. Embo J. 1987, 6, 2987–2995.
[96] Milligan, G., Kostenis, E., Heterotrimeric G-proteins: a short [115] Nurse, P., Masui, Y., Hartwell, L., Understanding the cell
history. Brit. J. Pharmacol. 2006, 147, S46–S55. cycle. Nat. Med. 1998, 4, 1103–1106.
[97] Dulbecco, R., Vogt, M., Plaque formation and isolation of [116] Lee, M. G., Nurse, P., Complementation used to clone a
pure lines with poliomyelitis viruses. J. Exp. Med. 1954, 99, human homolog of the fission yeast-cell cycle control gene
167–182. cdc2. Nature 1987, 327, 31–35.
[98] Laemmli, U. K., Cleavage of structural proteins during as- [117] Gautier, J., Norbury, C., Lohka, M., Nurse, P. et al., Puri-
sembly of head of bacteriophage-T4. Nature 1970, 227, fied maturation-promoting factor contains the product of
680–685. a Xenopus homolog of the fission yeast-cell cycle control
[99] Kohler, G., Milstein, C., Continuous cultures of fused cells gene cdc2+. Cell 1988, 54, 433–439.
secreting antibody of predefined specificity. Nature 1975, [118] Labbe, J. C., Lee, M. G., Nurse, P., Picard, A. et al., Activa-
256, 495–497. tion at M-phase of a protein-kinase encoded by a starfish
[100] Gubler, U., Hoffman, B. J., A simple and very efficient homolog of the cell-cycle control gene cdc2+. Nature 1988,
method for generating cDNA libraries. Gene 1983, 25, 263– 335, 251–254.
269. [119] Gautier, J., Minshull, J., Lohka, M., Glotzer, M. et al., Cy-
[101] Okayama, H., Berg, P., High-efficiency cloning of full-length clin is a component of maturation-promoting factor from
cDNA. Mol. Cell. Biol. 1982, 2, 161–170. Xenopus. Cell 1990, 60, 487–494.
[102] Southern, E. M., Detection of specific sequences among [120] Pines, J., Hunter, T., Isolation of a human cyclin CDNA - evi-
DNA fragments separated by gel-electrophoresis. J. Mol. dence for cyclin messenger-RNA and protein-regulation in
Biol. 1975, 98, 503–517. the cell-cycle and for interaction with p34-cdc2. Cell 1989,
[103] Sanger, F., Nicklen, S., Coulson, A. R., DNA sequencing 58, 833–846.
with chain-terminating inhibitors. Proc. Natl. Acad. Sci. [121] Solomon, M. J., Glotzer, M., Lee, T. H., Philippe, M. et al.,
USA 1977, 74, 5463–5467. Cyclin activation of p34cdc2. Cell 1990, 63, 1013–1024.
[104] Meselson, M., Yuan, R., DNA restriction enzyme from [122] Crawford, L. V., Lane, D. P., Denhardt, D. T., Harlow, E. E.
E. coli. Nature 1968, 217, 1110–1114. et al., Characterization of the complex between sv40 large
[105] Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J. et al., T-antigen and the 53k host protein in transformed mouse
Primer-directed enzymatic amplification of DNA with a cells. Cold Spring Harb. Symp. Quant. Biol. 1979, 44, 179–
thermostable DNA-polymerase. Science 1988, 239, 487– 187.
491. [123] Lane, D. P., Crawford, L. V., T-antigen is bound to a host
[106] Wagner, R. P., Rudolph Virchow and the genetic basis of protein in SV40-transformed cells. Nature 1979, 278, 261–
somatic ecology. Genetics 1999, 151, 917–920. 263.

[107] Hartwell, L. H., Culotti, J., Pringle, J. R., Reid, B. J., Genetic- [124] Brugge, J. S., Erikson, R. L., Identification of a
control of cell-division cycle in yeast. Science 1974, 183, transformation-specific antigen-induced by an avian-
46–51. sarcoma virus. Nature 1977, 269, 346–348.

[108] Nurse, P., Thuriaux, P., Nasmyth, K., Genetic-control of [125] Levinson, A. D., Oppermann, H., Levintow, L., Varmus,
cell-division cycle in fission yeast Schizosaccharomyces H. E., et al., Evidence that transforming gene of avian-
pombe. Mol. Gen. Genet. 1976, 146, 167–178. sarcoma virus encodes a protein-kinase associated with
a phosphoprotein. Cell 1978, 15, 561–572.
[109] Durkacz, B., Beach, D., Hayles, J., Nurse, P., The fission
yeast-cell cycle control gene cdc2 - structure of the cdc2 [126] Collett, M. S., Erikson, R. L., Protein-kinase activity associ-
region. Mol. Gen. Genet. 1985, 201, 543–545. ated with avian-sarcoma virus src gene product. Proc. Natl.
Acad. Sci. USA 1978, 75, 2021–2024.
[110] Simanis, V., Nurse, P., The cell-cycle control gene
cdc2+ of fission yeast encodes a protein-kinase po- [127] Eckhart, W., Hutchinson, M. A., Hunter, T., Activity phos-
tentially regulated by phosphorylation. Cell 1986, 45, phorylating tyrosine in polyoma T-antigen immunoprecip-
261–268. itates. Cell 1979, 18, 925–933.


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
1494 P. Braun and A.-C. Gingras Proteomics 2012, 12, 1478–1498

[128] Ushiro, H., Cohen, S., Identification of phosphotyrosine as modulation of AMPA/kainate receptors on hippocampal-
a product of epidermal growth factor-activated protein- neurons. Nature 1994, 368, 853–856.
kinase in A431 cell membranes. J. Biol. Chem. 1980, 255, [145] Mor, A., Philips, M. R., Compartmentalized Ras/MAPK sig-
8363–8365. naling. Annual Review of Immunology, Annual Reviews,
[129] Konopka, J. B., Watanabe, S. M., Witte, O. N., An alter- Palo Alto 2006, pp. 771–800.
ation of the human c-abl protein in K562 leukemia-cells [146] Jacob, F., Ullmann, A., Monod, J., Délétions fusionnant
unmasks associated tyrosine kinase-activity. Cell 1984, 37, l’opéron lactose et un opéron purine chez Escherichia coli.
1035–1042. J. Mol. Biol. 1965, 13, 704–719.
[130] Gribskov, M., McLachlan, A. D., Eisenberg, D., Profile anal-
[147] Shuman, H. A., Silhavy, T. J., Beckwith, J. R., Labeling of
ysis - detection of distantly related proteins. Proc. Natl.
proteins with beta-galactosidase by gene fusion - iden-
Acad. Sci. USA 1987, 84, 4355–4358.
tification of a cytoplasmic membrane component of the
[131] Altschul, S. F., Gish, W., Miller, W., Myers, E. W., et al., Escherichia coli maltose transport-system. J. Biol. Chem.
Basic local alignment search tool. J. Mol. Biol. 1990, 215, 1980, 255, 168–174.
403–410.
[148] Germino, J., Bastia, D., Rapid purification of a cloned gene-
[132] Sadowski, I., Stone, J. C., Pawson, T., A noncatalytic do- product by genetic fusion and site-specific proteolysis.
main conserved among cytoplasmic protein-tyrosine ki- Proc. Natl. Acad. Sci. USA-Biolog. Sci. 1984, 81, 4692–
nases modifies the kinase function and transforming activ- 4696.
ity of Fujinami sarcoma-virus p130gag-fps. Mol. Cell. Biol.
[149] Smith, D. B., Johnson, K. S., Single-step purification of
1986, 6, 4396–4408.
polypeptides expressed in escherichia-coli as fusions with
[133] Anderson, D., Koch, C. A., Grey, L., Ellis, C. et al., Binding of glutathione s-transferase. Gene 1988, 67, 31–40.
SH2 domains of Phospholipase-C-gamma-1, GAP, and Src
[150] Hammarberg, B., Nygren, P. A., Holmgren, E., Elmblad,
to activated growth-factor receptors. Science 1990, 250,
A. et al., Dual affinity fusion approach and its use to
979–982.
express recombinant human insulin-like growth factor-ii.
[134] Schlessinger, J., Cell signaling by receptor tyrosine ki- Proc. Natl. Acad. Sci. USA 1989, 86, 4367–4371.
nases. Cell 2000, 103, 211–225.
[151] Rigaut, G., Shevchenko, A., Rutz, B., Wilm, M. et al., A
[135] Pawson, T., Protein modules and signaling networks. generic protein purification method for protein complex
Nature 1995, 373, 573–580. characterization and proteome exploration. Nat. Biotech-
[136] Whyte, P., Buchkovich, K. J., Horowitz, J. M., Friend, nol. 1999, 17, 1030–1032.
S. H. et al., Association between an oncogene and an
[152] Puig, O., Caspary, F., Rigaut, G., Rutz, B. et al., The tandem
anti-oncogene - the adenovirus E1A proteins bind to
affinity purification (TAP) method: a general procedure
the retinoblastoma gene-product. Nature 1988, 334, 124–
of protein complex purification. Methods 2001, 24, 218–
129.
229.
[137] Baker, S. J., Fearon, E. R., Nigro, J. M., Hamilton, S. R.
[153] Nilsson, J., Stahl, S., Lundeberg, J., Uhlen, M. et al., Affin-
et al., Chromosome-17 deletions and p53 gene-mutations
ity fusion strategies for detection, purification, and immo-
in colorectal carcinomas. Science 1989, 244, 217–221.
bilization of recombinant proteins. Protein Expres. Purif.
[138] Oltvai, Z. N., Milliman, C. L., Korsmeyer, S. J., Bcl-2 het- 1997, 11, 1–16.
erodimerizes in vivo with a conserved homolog, bax, that
[154] Niman, H. L., Houghten, R. A., Walker, L. E., Reisfeld, R. A.
accelerates programmed cell-death. Cell 1993, 74, 609–619.
et al., Generation of protein-reactive antibodies by short
[139] Colledge, M., Scott, J. D., AKAPs: from structure to func- peptides is an event of high-frequency - implications for
tion. Trends in Cell Biol. 1999, 9, 216–221. the structural basis of immune recognition. Proc. Natl.
[140] Taylor, S. S., Buechler, J. A., Yonemoto, W., cAMP- Acad. Sci. USA-Biolog. Sci. 1983, 80, 4949–4953.
dependent protein-kinase - framework for a diverse family [155] Edman, P., Begg, G., A protein sequenator. Eur. J. Biochem.
of regulatory enzymes. Annu. Rev. Biochem. 1990, 59, 971– 1967, 1, 80–91.
1005.
[156] Karas, M., Hillenkamp, F., Laser desorption ionization of
[141] Smith, F. D., Scott, J. D., Anchored cAMP signaling: onward proteins with molecular masses exceeding 10000 daltons.
and upward - a short history of compartmentalized cAMP Anal. Chem. 1988, 60, 2299–2301.
signal transduction. Eur. J. Cell Biol. 2006, 85, 585–592.
[157] Fenn, J. B., Mann, M., Meng, C. K., Wong, S. F. et al.,
[142] Buxton, I. L. O., Brunton, L. L., Compartments of cyclic- Electrospray ionization for mass-spectrometry of large
AMP and protein-kinase in mammalian cardiomyocytes. biomolecules. Science 1989, 246, 64–71.
J. Biol. Chem. 1983, 258, 233–239.
[158] Louris, J. N., Wright, L. G., Cooks, R. G., Schoen, A. E., New
[143] Bregman, D. B., Bhattacharyya, N., Rubin, C. S., High- scan modes accessed with a hybrid mass-spectrometer.
affinity binding-protein for the regulatory subunit of camp- Anal. Chem. 1985, 57, 2918–2924.
dependent protein kinase II-b - cloning, characterization,
[159] Aebersold, R., Mann, M., Mass spectrometry-based pro-
and expression of cdnas for rat-brain p150. J. Biol. Chem.
teomics. Nature 2003, 422, 198–207.
1989, 264, 4648–4656.
[160] Fields, S., Song, O., A novel genetic system to detect
[144] Rosenmund, C., Carr, D. W., Bergeson, S. E., Nilaver,
protein-protein interactions. Nature 1989, 340, 245–246.
G. et al., Anchoring of protein-kinase-A is required for


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
Proteomics 2012, 12, 1478–1498 1495

[161] Vojtek, A. B., Hollenberg, S. M., Cooper, J. A., Mammalian proteomics applications - improved expression by addi-
ras interacts directly with the serine threonine kinase raf. tion of splice sites. BMC Biotechnol. 2006, 6, 13. DOI:
Cell 1993, 74, 205–214. 10.1186/1472-6750-6-13.
[162] Bartel, P. L., Roecklein, J. A., SenGupta, D., Fields, S., A [178] Park, J., Labaer, J., UNIT 3.20 Recombinational cloning.
protein linkage map of Escherichia coli bacteriophage t7. Curr. Protocols Mol. Biol./edited by Frederick M. Ausubel
Nat. Genet. 1996, 12, 72–77. . . . [et al.] 2006, Chapter 3.
[163] Petschnigg, J., Moe, O. W., Stagljar, I., Using yeast as a [179] Nilsson, S., Svedberg, T. M., Pettersson, J., Bjorefors, T. F.
model to study membrane proteins. Curr. Opin. Nephrol. et al., Evaluations of the stability of sheathless electrospray
Hy. 2011, 20, 425–432. ionization mass spectrometry emitters using electrochem-
[164] Michnick, S. W., Ear, P. H., Landry, C., Malleshaiah, M. K. ical techniques. Anal. Chem. 2001, 73, 4607–4616.
et al., in: Weissman, J., Guthrie, C., Fink, G. R. (Eds.), Meth- [180] Oliver, S., Guilt-by-association goes global. Nature 2000,
ods in Enzymology, Vol 470: Guide to Yeast Genetics: Func- 403, 601–603.
tional Genomics, Proteomics, and Other Systems Analysis, [181] Uetz, P., Giot, L., Cagney, G., Mansfield, T., A compre-
2nd Edition, Elsevier Academic Press Inc., San Diego 2010, hensive analysis of protein-protein interactions in Saccha-
pp. 335–368. romyces cerevisiae. Nature 2000, 403, 623–627.
[165] Lievens, S., Eyckerman, S., Lemmens, I., Tavernier, J., [182] Walhout, A. J., Sordella, R., Lu, X., Hartley, J. L. et al.,
Large-scale protein interactome mapping: strategies and Protein interaction mapping in C. elegans using proteins
opportunities. Exp. Rev. Proteomics 2010, 7, 679–690. involved in vulval development. Science 2000, 287, 116–
[166] Vidal, M., Legrain, P., Yeast forward and reverse ‘n’-hybrid 122.
systems. Nucleic Acids Res. 1999, 27, 919–929. [183] Ito, T., Tashiro, K., Muta, S., Ozawa, R. et al., Toward
[167] Braun, P., Interactome mapping for analysis of com- a protein-protein interaction map of the budding yeast:
plex phenotypes: insights from benchmarking bi- a comprehensive system to examine two-hybrid interac-
nary interaction assays. Proteomics 2012, 12, doi: tions in all possible combinations between the yeast pro-
10.1002/pmic.201100598. teins. Proc. Natl. Acad. Sci. USA 2000, 97, 1143–1147.
[168] Roberts, L., Davenport, R. J., Pennisi, E., Marshall, E., A [184] Ito, T., Chiba, T., Ozawa, R., Yoshida, M. et al., A compre-
history of the human genome project. Science (New York, hensive two-hybrid analysis to explore the yeast protein
N.Y.) 2001, 291, 1195. interactome. Proc. Natl. Acad. Sci. USA 2001, 98, 4569–
[169] Walhout, A. J., Temple, G. F., Brasch, M. A., Hartley, J. 4574.
L. et al., Gateway recombinational cloning: application to [185] Gavin, A., Bosche, M., Krause, R., Grandi, P., Functional
the cloning of large numbers of open reading frames or organization of the yeast proteome by systematic analysis
ORFeomes. Methods Enzymol. 2000, 328, 575–592. of protein complexes. Nature 2002, 415, 141–147.
[170] Reboul, J., Vaglio, P., Rual, J. F., Lamesch, P. et al., [186] Ho, Y., Gruhler, A., Heilbut, A., Bader, G., Systematic identi-
C. elegans ORFeome version 1.1: experimental verification fication of protein complexes in Saccharomyces cerevisiae
of the genome annotation and resource for proteome-scale by mass spectrometry. Nature 2002, 415, 180–183.
protein expression. Nat. Genet. 2003, 34, 35–41. [187] Gavin, A. C., Aloy, P., Grandi, P., Krause, R. et al., Proteome
[171] Lamesch, P., Li, N., Milstein, S., Fan, C. et al., HORFeome survey reveals modularity of the yeast cell machinery. Na-
v3.1: a resource of human open reading frames represent- ture 2006, 440, 631–636.
ing over 10,000 human genes. Genomics 2007, 89, 307– [188] Krogan, N. J., Cagney, G., Yu, H., Zhong, G. et al., Global
315. landscape of protein complexes in the yeast Saccha-
[172] Hilson, P., Cloned sequence repertoires for small- and romyces cerevisiae. Nature 2006, 440, 637–643.
large-scale biology. Trends Plant Sci. 2006, 11, 133–141. [189] Giot, L., Bader, J., Brouwer, C., Chaudhuri, A., A protein in-
[173] Yamada, K., Lim, J., Dale, J. M., Chen, H. M. et al., Empir- teraction map of Drosophila melanogaster. Science 2003,
ical analysis of transcriptional activity in the Arabidopsis 302, 1727–1736.
genome. Science 2003, 302, 842–846. [190] Li, S., Armstrong, C. M., Bertin, N., Ge, H. et al., A map
[174] Rolfs, A., Hu, Y., Ebert, L., Hoffmann, D. et al., A biomedi- of the interactome network of the metazoan C. elegans.
cally enriched collection of 7000 human ORF clones. PLoS Science 2004, 303, 540–543.
One 2008, 3, e1528. DOI: 10.1371/journal.pone.0001528 [191] Rual, J. F., Venkatesan, K., Hao, T., Hirozane-Kishikawa,
[175] Yang, X., Boehm, J. S., Yang, X., Salehi-Ashtiani, K. et al., A T. et al., Towards a proteome-scale map of the human
public genome-scale lentiviral expression library of human protein-protein interaction network. Nature 2005, 437,
ORFs. Nat. Methods 2011, 8, U659–U680. 1173–1178.
[176] Yu, C., Wan, K. H., Hammonds, A. S., Stapleton, M. et al., [192] Stelzl, U., Worm, U., Lalowski, M., Haenig, C. et al., A hu-
in: Wu, C. J. (Ed.), Protein Microarray for Disease Analysis: man protein-protein interaction network: a resource for
Methods and Protocols, Humana Press Inc., Totowa, NJ annotating the proteome. Cell 2005, 122, 957–968.
2011, pp. 257–272. [193] Arabidopsis Interactome Mapping Consortium, Evidence
[177] Colwill, K., Wells, C. D., Elder, K., Goudreault, M. et al., for network evolution in an Arabidopsis interactome map.
Modification of the creator recombination system for Science 2011, 333, 601–607.


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
1496 P. Braun and A.-C. Gingras Proteomics 2012, 12, 1478–1498

[194] Sato, S., Shimoda, Y., Muraki, A., Kohara, M. et al., A large- [210] Schwikowski, B., Uetz, P., Fields, S., A network of protein-
scale protein protein interaction analysis in Synechocystis protein interactions in yeast. Nat. Biotechnol. 2000, 18,
sp. Pcc6803. DNA Res. 2007, 14, 207–216. 1257–1261.
[195] Shimoda, Y., Shinpo, S., Kohara, M., Nakamura, Y. et al., A [211] Vidal, M., Cusick, M. E., Barabasi, A.-L., Interactome net-
large scale analysis of protein-protein interactions in the works and human disease. Cell 2011, 144, 986–998.
nitrogen-fixing bacterium Mesorhizobium loti. DNA Res. [212] Barabasi, A. L., Bonabeau, E., Scale-free networks. Sci. Am.
2008, 15, 13–23. 2003, 288, 60–69.
[196] Tarassov, K., Messier, V., Landry, C. R., Radinovic, S. et al., [213] Albert, R., Jeong, H., Barabasi, A. L., Error and attack toler-
An in vivo map of the yeast protein interactome. Science ance of complex networks. Nature 2000, 406, 378–382.
2008, 320, 1465–1470.
[214] Calderwood, M. A., Venkatesan, K., Xing, L., Chase, M. R.
[197] Miller, J. P., Lo, R. S., Ben-Hur, A., Desmarais, C. et al., et al., Epstein-Barr virus and virus human protein interac-
Large-scale identification of yeast integral membrane pro- tion maps. Proc. Natl. Acad. Sci. USA 2007, 104, 7606–7611.
tein interactions. Proc. Natl. Acad. Sci. USA 2005, 102,
[215] Mukhtar, M. S., Carvunis, A.-R., Dreze, M., Epple, P. et al.,
12123–12128.
Independently evolved virulence effectors converge onto
[198] Lalonde, S., Sero, A., Pratelli, R. J., Pilot, G. et al., A hubs in a plant immune system network. Science 2011,
membrane protein/signaling protein interaction network 333, 596–601.
for Arabidopsis version ampv2. Front. Physiol. 2010, 1.
[216] Friedel, C. C., Haas, J., Virus-host interactomes and global
DOI: 10.3389/fphys.2010.00024.
models of virus-infected cells. Trends in Microbiol. 2011,
[199] Huh, W. K., Falvo, J. V., Gerke, L. C., Carroll, A. S. et al., 19, 501–508.
Global analysis of protein localization in budding yeast.
[217] Yang, H., Ke, Y., Wang, J., Tan, Y. et al., Insight into bacte-
Nature 2003, 425, 686–691.
rial virulence mechanisms against host immune response
[200] Ghaemmaghami, S., Huh, W., Bower, K., Howson, R. W. via the Yersinia pestis-human protein-protein interaction
et al., Global analysis of protein expression in yeast. Nature network. Infect. Immun. 2011, 79, 4413–4424.
2003, 425, 737–741.
[218] Navratil, V., de Chassey, B., Combe, C. R., Lotteau, V.,
[201] Winzeler, E. A., Shoemaker, D. D., Astromoff, A., Liang, When the human viral infectome and diseasome networks
H. et al., Functional characterization of the S. cerevisiae collide: towards a systems biology platform for the ae-
genome by gene deletion and parallel analysis. Science tiology of human diseases. BMC Syst. Biol. 2011, 5, 13.
1999, 285, 901–906. DOI:10.1186/1752-0509-5-13.
[202] Giaever, G., Chu, A. M., Ni, L., Connelly, C. et al., Functional [219] Yu, H., Braun, P., Yildirim, M. A., Lemmens, I. et al., High-
profiling of the saccharomyces cerevisiae genome. Nature quality binary protein interaction map of the yeast interac-
2002, 418, 387–391. tome network. Science 2008, 322, 104–110.
[203] Tong, A. H. Y., Lesage, G., Bader, G. D., Ding, H. M. et al., [220] Goh, K. I., Cusick, M. E., Valle, D., Childs, B. et al., The
Global mapping of the yeast genetic interaction network. human disease network. Proc. Natl. Acad. Sci. USA 2007,
Science 2004, 303, 808–813. 104, 8685–8690.
[204] Zhu, H., Klemic, J. F., Chang, S., Bertone, P. et al., Analysis [221] Lim, J., Hao, T., Shaw, C., Patel, A. J. et al., A protein-
of yeast protein kinases using protein chips. Nat. Genet. protein interaction network for human inherited ataxias
2000, 26, 283–289. and disorders of Purkinje cell degeneration. Cell 2006, 125,
[205] Ptacek, J., Devgan, G., Michaud, G., Zhu, H. et al., Global 801–814.
analysis of protein phosphorylation in yeast. Nature 2005, [222] Soler-Lopez, M., Zanzoni, A., Lluis, R., Stelzl, U. et al., In-
438, 679–684. teractome mapping suggests new mechanistic details un-
[206] Ge, H., Liu, Z. H., Church, G. M., Vidal, M., Correlation derlying Alzheimer’s disease. Genome Res. 2011, 21, 364–
between transcriptome and interactome mapping data 376.
from saccharomyces cerevisiae. Nat. Genet. 2001, 29, 482– [223] Pujana, M. A., Han, J. D., Starita, L. M., Stevens, K. N. et al.,
486. Network modeling links breast cancer susceptibility and
[207] Von Mering, C., Jensen, L. J., Snel, B., Hooper, S. D. centrosome dysfunction. Nat. Genet. 2007, 39, 1338–1349.
et al., String: known and predicted protein-protein asso- [224] Lage, K., Karlberg, E. O., Storling, Z. M., Olason, P. I. et al.,
ciations, integrated and transferred across organisms. Nu- A human phenome-interactome network of protein com-
cleic Acids Res. 2005, 33, D433–D437. plexes implicated in genetic disorders. Nat. Biotech. 2007,
[208] Jensen, L. J., Kuhn, M., Stark, M., Chaffron, S. et al., String 25, 309–316.
8-a global view on proteins and their functional interac- [225] Breitkreutz, A., Choi, H., Sharom, J. R., Boucher, L. et al., A
tions in 630 organisms. Nucleic Acids Res. 2009, 37, D412– global protein kinase and phosphatase interaction network
D416. in yeast. Science 2010, 328, 1043–1046.
[209] Jansen, R., Yu, H. Y., Greenbaum, D., Kluger, Y. et al., A [226] Ulitsky, I., Shlomi, T., Kupiec, M., Shamir, R., From E-maps
Bayesian networks approach for predicting protein-protein to module maps: dissecting quantitative genetic interac-
interactions from genomic data. Science 2003, 302, 449– tions using physical interactions. Mol. Syst. Biol. 2008, 4,
453. 209.


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
Proteomics 2012, 12, 1478–1498 1497

[227] Bandyopadhyay S., Kelley R., Krogan N. J., Ideker T., [244] Lavallee-Adam, M., Cloutier, P., Coulombe, B., Blanchette,
Functional maps of protein complexes from quantitative M., Modeling contaminants in AP-MS/MS experiments.
genetic interaction data. PLoS Computat. Biol. 2008, 4, J. Proteome Res. 2011, 10, 886–895.
e1000065. doi:10.1371/journal.pcbi.1000065. [245] Liu, G., Zhang, J., Larsen, B., Stark, C. et al., Prohits: inte-
[228] Hannum, G., Srivas, R., Guenole, A., van Attikum, H. et al., grated software for mass spectrometry-based interaction
Genome-wide association data reveal a global map of ge- proteomics. Nat. Biotechnol. 2010, 28, 1015–1017.
netic interactions among protein complexes. PLoS Genet- [246] Choi, H., Larsen, B., Lin, Z.-Y., Breitkreutz, A. et al., Saint:
ics 2009, 5. probabilistic scoring of affinity purification-mass spec-
[229] Costanzo, M., Baryshnikova, A., Bellay, J., Kim, Y. et al., The trometry data. Nat. Methods 2011, 8, U70–U100.
genetic landscape of a cell. Science 2010, 327, 425–431. [247] Ranish, J. A., Hahn, S., Lu, Y., Yi, E. C. et al., Identification of
[230] Michaut, M., Baryshnikova, A., Costanzo, M., Myers, C. L. TFB5, a new component of general transcription and DNA
et al., Protein complexes are central in the yeast genetic repair factor iih. Nat. Genet. 2004, 36, 707–713.
landscape. PLoS Computat. Biol. 2011, 7, e1001092. [248] Trinkle-Mulcahy, L., Andersen, J., Lam, Y. W., Moorhead,
[231] Akula, N., Baranova, A., Seto, D., Solka, J. et al., A network- G. et al., Repo-man recruits pp1 gamma to chromatin and
based approach to prioritize results from genome-wide as- is essential for cell viability. J. Cell Biol. 2006, 172, 679–
sociation studies. PLoS One 2011, 6, e24220. 692.
[232] Rossin, E. J., Lage, K., Raychaudhuri, S., Xavier, R. J. [249] Selbach, M., Mann, M., Protein interaction screening by
et al., Proteins encoded in genomic regions associated with quantitative immunoprecipitation combined with knock-
immune-mediated disease physically interact and suggest down (QUICK). Nat. Methods 2006, 3, 981–983.
underlying biology. PLoS Genet. 2011, 7, e1001273. [250] Hubner, N. C., Bird, A. W., Cox, J., Splettstoesser, B.
[233] Taylor, I. W., Linding, R., Warde-Farley, D., Liu, Y. et al., Dy- et al., Quantitative proteomics combined with bac trans-
namic modularity in protein interaction networks predicts geneomics reveals in vivo protein interactions. J. Cell Biol.
breast cancer outcome. Nat. Biotechnol. 2009, 27, 199–204. 2010, 189, 739–754.
[234] Chuang, H.-Y., Lee, E., Liu, Y.-T., Lee, D. et al., Network- [251] Uhlens, M., Bjorling, E., Agaton, C., Szigyarto, C. A. et al., A
based classification of breast cancer metastasis. Mol. Syst. human protein atlas for normal and cancer tissues based
Biol. 2007, 3, 140. on antibody proteomics. Mol. Cell. Proteomics 2005, 4,
[235] Barabasi, A.-L., Gulbahce, N., Loscalzo, J., Network 1920–1932.
medicine: a network-based approach to human disease. [252] Persson, A., Hober, S., Uhlen, M., A human protein atlas
Nat. Rev. Genet. 2011, 12, 56–68. based on antibody proteomics. Curr. Opin. Mol. Therap.
[236] Venkatesan, K., Rual, J. F., Vazquez, A., Stelzl, U. et al., An 2006, 8, 185–190.
empirical framework for binary interactome mapping. Nat. [253] Colwill, K., Graslund, S., Renewable prot binder working,
Methods 2009, 6, 83–90. G., A roadmap to generate renewable protein binders
[237] Braun, P., Tasan, M., Dreze, M., Barrios-Rodiles, M. et al., to the human proteome. Nat. Methods 2011, 8, U551–
An experimentally derived confidence score for binary U552.
protein-protein interactions. Nat. Methods 2009, 6, 91–97. [254] Taussig, M. J., Stoevesandt, O., Borrebaeck, C. A. K., Brad-
[238] Chen, Y. C., Rajagopala, S. V., Stellberger, T., Uetz, P., Ex- bury, A. R. et al., Proteomebinders: planning a European
haustive benchmarking of the yeast two-hybrid system. resource of affinity reagents for analysis of the human pro-
Nat. Methods 2010, 7, 667–668; author reply 668. teome. Nat. Methods 2007, 4, 13–17.

[239] Simonis, N., Rual, J. F., Carvunis, A. R., Tasan, M. et al., [255] Malovannaya, A., Lanz, R. B., Jung, S. Y., Bulynko, Y. et al.,
Empirically controlled mapping of the Caenorhabditis ele- Analysis of the human endogenous coregulator complex-
gans protein–protein interactome network. Nat. Methods ome. Cell 2011, 145, 787–799.
2009, 47–54. [256] Reichert, J. M., Monoclonal antibodies as innovative ther-
[240] Stumpf, M. P. H., Thorne, T., de Silva, E., Stewart, R. et al., apeutics. Curr. Pharm. Biotechnol. 2008, 9, 423–430.
Estimating the size of the human interactome. Proc. Natl. [257] Reichert, J. M., Rosensweig, C. J., Faden, L. B., Dewitz,
Acad. Sci. USA 2008, 105, 6959–6964. M. C., Monoclonal antibody successes in the clinic. Nat.
[241] Sambourg, L., Thierry-Mieg, N., New insights into protein- Biotechnol. 2005, 23, 1073–1078.
protein interaction data lead to increased estimates of the [258] Ravelli, R. B. G., Gigant, B., Curmi, P. A., Jourdain, I.
S. cerevisiae interactome size. BMC Bioinform. 2010, 11, et al., Insight into tubulin regulation from a complex with
605. colchicine and a stathmin-like domain. Nature 2004, 428,
[242] Sowa, M. E., Bennett, E. J., Gygi, S. P., Harper, J. W., Defin- 198–202.
ing the human deubiquitinating enzyme interaction land- [259] Pommier, Y., Marchand, C., Interfacial inhibitors: targeting
scape. Cell 2009, 138, 389–403. macromolecular complexes. Nat. Rev. Drug Discov. 2012,
[243] Sardiu, M. E., Cai, Y., Jin, J., Swanson, S. K. et al., Prob- 11, 25–36.
abilistic assembly of human protein interaction networks [260] Wells, J. A., McClendon, C. L., Reaching for high-hanging
from label-free quantitative proteomics. Proc. Natl. Acad. fruit in drug discovery at protein-protein interfaces. Nature
Sci. USA 2008, 105, 1454–1459. 2007, 450, 1001–1009.


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
1498 P. Braun and A.-C. Gingras Proteomics 2012, 12, 1478–1498

[261] Arkin, M. R., Wells, J. A., Small-molecule inhibitors [266] Barabasi, A.-L., The network takeover. Nat. Phys. 2012, 8,
of protein-protein interactions: progressing towards 14–16.
the dream. Nat. Rev. Drug Discov. 2004, 3, 301– [267] Chuang, H.-Y., Hofree, M., Ideker, T., A decade of systems
317. biology. Ann. Rev. Cell Dev. Biol. 2010, 26, 721–744.
[262] Vidal, M., Braun, P., Chen, E., Boeke, J. D. et al., Ge- [268] Zanzoni, A., Soler-Lopez, M., Aloy, P., A network medicine
netic characterization of a mammalian protein-protein in- approach to human disease. FEBS Lett. 2009, 583, 1759–
teraction domain by using a yeast reverse two-hybrid 1765.
system. Proc. Natl. Acad. Sci. USA 1996, 93, 10321–
[269] Zhong, Q., Simonis, N., Li, Q. R., Charloteaux, B. et al.,
10326.
Edgetic perturbation models of human inherited disorders.
[263] Dreze, M., Charloteaux, B., Milstein, S., Vidalain, P. O. et al., Mol. Syst. Biol. 2009, 5, 321.
‘Edgetic’ perturbation of a C. elegans BCL-2 ortholog. Nat.
[270] Gammie, A. E., Erdeniz, N., Beaver, J., Devlin, B. et al.,
Methods 2009, 6, 843–849.
Functional characterization of pathogenic human MSH2
[264] Patel, S., Player, M. R., Small-molecule inhibitors of the missense mutations in Saccharomyces cerevisiae. Genet-
p53-HDM2 interaction for the treatment of cancer. Expert ics 2007, 177, 707–721.
Opin. Invest. Drug. 2008, 17, 1865–1882.
[271] Yamamoto, Y., Hirai, T., Yamamoto, E., Kawamura, M. et al.,
[265] Barabasi, A. L., Oltvai, Z. N., Network biology: understand- A rice GID1 suppressor mutant reveals that gibberellin is
ing the cell’s functional organization. Nat. Rev. Genet. 2004, not always required for interaction between its receptor,
5, 101–113. GID1, and DELLA proteins. Plant Cell 2010, 22, 3589–3602.


C 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com

You might also like