Analyzing protein–protein interactions by quantitative mass spectrometry
Florian E. Paul, Fabian Hosp, Matthias Selbach⇑
Cell Signaling and Mass Spectrometry Group, Max Delbrück Center for Molecular Medicine, Berlin, Germany
a r t i c l e i n f o
Available online 5 March 2011
a b s t r a c t
Since most cellular processes depend on interactions between proteins, information about protein–
protein interactions (PPIs) provide valuable insights into protein function. Over the last years, quantita-
tive affinity purification followed by mass spectrometry (q-AP-MS) has become a powerful approach to
investigate PPIs in an unbiased manner. In q-AP-MS the protein of interest is biochemically enriched
together with its interaction partners. In parallel, a control experiment is performed to control for
non-specific binding. Quantitative mass spectrometry is then employed to compare protein levels in both
samples and to exclude non-specific contaminants. Here, we provide two detailed q-AP-MS protocols for
pull-downs with immobilized bait proteins or transient transfection of tagged expression constructs. We
discuss benefits and limitations of q-AP-MS and highlight critical parameters that need to be considered.
The protocols and background information presented here allow the reader to adapt the generic q-AP-MS
strategy for a wide range of biological questions.
? 2011 Elsevier Inc. All rights reserved.
Most biological processes require direct physical interactions
between proteins. While some interactions are binary, other cellu-
lar events involve large multi-protein complexes. Identifying and
characterizing protein–protein interactions (PPIs) is crucial to gain
molecular insights into cell function and physiology. Importantly,
PPIs are frequently regulated in response to a specific stimulus or
cell state [1,2]. It is therefore important to identify binding part-
ners of a protein in its specific cellular context. An ideal system
for detecting PPIs would take the cellular background into account,
work with endogenous levels of the protein of interest and have
both a high sensitivity (i.e. low false-negative rate) and specificity
(i.e. low false-positive rate).
One of the most popular methods of studying PPIs is the yeast-
two-hybrid (Y2H) approach . A great advantage of this system is
that it is scalable and can be used to identify many PPIs in rela-
tively short time. However, the Y2H method provides only a static
picture and cannot yield immediate clues about the cellular pro-
cesses that convert genetic information into complex phenotypes
. Other disadvantages are caused by using yeast as a heterolo-
gous system. For example, mammalian proteins expressed in yeast
may not carry all post-translational modifications relevant for their
function. This is especially problematic since PPIs are often regu-
lated by reversible modifications such as phosphorylation [1,5].
These and other factors contribute to the notorious high false-
positive and false-negative rate of the assay.
An attractive alternative to Y2H is affinity purification followed
by mass spectrometry or AP-MS (see  for an excellent review).
ically from an appropriate biological sample. In most cases tissue
culture cells are employed but in vivo samples such as whole organs
can also be used. After purification, mass spectrometry-based pro-
teomics [7,8] identifies the proteins in the sample. This list of iden-
tified proteins is expected to contain the protein of interest and its
rived from their native cellular environment that carry all relevant
post-translational modifications. The method reveals the composi-
tion of entire protein complexes and thus potentially the function
of large molecular machines. When combined with quantification,
the method can also uncover dynamic changes in PPIs and thereby
directly provide information about cell signaling.
The biggest challenge in AP-MS experiments is to distinguish
between true interaction partners and co-purifying contaminants.
One way to alleviate this problem is tandem affinity purification
(TAP). In this method, tagged bait proteins are purified by two suc-
cessive purification steps in order to reduce the non-specific back-
ground binders . Although the TAP method is still widely used it
has considerable disadvantages. Most importantly, due to the in-
creased sensitivity of mass spectrometers it is impossible to re-
move contaminants completely. Furthermore, two successive
stringent purification steps can also remove biologically important
but weak or substoichiometric interactors. Thus, TAP suffers from a
trade-off between specificity and sensitivity.
1046-2023/$ - see front matter ? 2011 Elsevier Inc. All rights reserved.
⇑Corresponding author. Address: Max Delbrück Center for Molecular Medicine,
Robert-Rössle-Str. 10, D-13092 Berlin, Germany. Fax: +49 30 9406 2394.
E-mail address: email@example.com (M. Selbach).
Methods 54 (2011) 387–395
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/ymeth
The best way of distinguishing true interaction partners from
non-specific contaminants available to date is to use quantitative
proteomics. This strategy is based on comparing the abundance
of proteins identified in the sample with a suitable control. The
general idea is to perform two experiments in parallel: In addition
to the affinity purification of the protein of interest a control exper-
iment is performed that is not expected to yield any interaction
partners. Quantitative mass spectrometry can then be used to com-
pare the abundance of proteins in both pull-downs. True interac-
tion partners are more abundant in the actual AP-MS sample
compared to the control. In contrast, non-specific contaminants
have a 1:1 ratio since they are equally abundant in both pull-
downs. In this way, quantification circumvents the trade-off be-
tween sensitivity and specificity and can confidently identify PPIs
even under low stringency conditions.
Quantitative AP-MS (q-AP-MS) experiments have been em-
ployed successfully in different ways to address a wide range of
biological questions [10–20]. A detailed discussion of this work is
beyond the scope of this manuscript, see  for an excellent re-
view. Here, we describe two strategies of how q-AP-MS can be used
to detect PPIs (Fig. 1). Both methods are generic and can be em-
ployed to a broad spectrum of biological questions. We provide
both step-by-step protocols and background information to allow
researchers adapting the workflow to their specific experimental
system. We begin with a discussion of general considerations
important for any q-AP-MS experiment.
2. General considerations
2.1. Exogenous or endogenous bait?
The gold-standard assay for PPIs is still the co-immunoprecipi-
tation (coIP) of untagged proteins at their endogenous level. The
only available screening method based on this assay is quantitative
immunoprecipitation combined with knock-down (QUICK) . In
QUICK, the protein of interest is knocked-down by RNA interfer-
ence in control cells but not in cells used for the real coIP. The pro-
tein of interest is then precipitated from both samples. The target
protein itself and its interaction partners are more abundant in
the real coIP compared to the control and can thus be identified.
QUICK assesses interactions between untagged endogenous pro-
teins at their normal cellular levels within the appropriate cell
type. It can therefore identify PPIs with very high confidence.
The disadvantage of QUICK is that antibodies are often not
available. Therefore, we describe two alternative strategies based
on tagged bait proteins that can be used even when bait-specific
antibodies are not available. In the first approach an exogenous
bait is coupled to a matrix to fish-out interaction partners from cell
lysates (Fig. 1A). This approach is very flexible and allows for the
application of different types of baits. For example, recombinant
proteins, synthetic peptides or other small molecules can be used
[10–12,15,17,19,20,22]. Another interesting feature is that the bait
can be used in modified forms to screen for modification-specific
interactions. For instance, differences in interaction partners of
tyrosine-phosphorylated peptides and their non-modified versions
can reveal phosphorylation-dependent interactions. In the exam-
ple presented here we use GDP- versus GTPcS-loaded states of
the small GTPase Cdc42. The experiment is therefore designed to
identify specific interaction partners of both the inactivated and
the activated GTPase.
In the second approach presented here the bait protein is ex-
pressed in cell lines by transient transfection with an expression
vector (Fig. 1B). Bait proteins are fused to a biochemical tag to facil-
itate purification. A wide range of different tags is available with
different advantages and disadvantages (see Table 1 for an over-
view and  for a detailed review). These tags differ in size,
ranging from short peptide motifs to proteins of several kDa. Tags
bind to different types of binding partners such as antibodies (e.g.
a-FLAG, a-HA), proteins (e.g. streptavidin, calmodulin), or small
molecules (e.g. biotin, glutathione). Smaller tags may be advanta-
geous since they are expected to interfere less with bait protein
function. On the other hand, larger tags can help to increase the
solubility of the bait protein which can facilitate subsequent puri-
fication. Another important factor to consider is the affinity of the
interaction between the tag and its binding partner: The interac-
tion must be strong enough to efficiently purify the bait from the
Pull-downs with exogenous or transfected baits have unique
strengths and weaknesses. Using exogenous baits uncouples bait
production from the interaction experiment. Therefore, large
amounts of cell lysates and bait proteins can be prepared in ad-
vance and used for many experiments. Even cells which cannot
be transfected efficiently can be used without restriction. However,
it should be kept in mind that this assay may erroneously detect
interactions among proteins that never co-localize in vivo. Trans-
fected bait proteins have the advantage that they are produced in
their native environment and in the correct subcellular location.
An important caveat here is that overexpression of the bait protein
might result in non-specific interactions. Expression systems based
on bacterial artificial chromosomes (BACs) circumvent this prob-
lem but are not available for all proteins of interest or specific vari-
ants . Irrespective of the method used it should always be kept
in mind that tagging may alter protein function. Therefore, interac-
tion partners should be verified by coIP with the endogenous pro-
teins and/or functional follow-ups.
2.2. Which method should be used for quantification?
q-AP-MS depends on reliable quantification of relative differ-
ences in protein abundance between in the pull-down of interest
and the control. Different methods allow for protein quantification
by mass spectrometry [24,25]. Some of these methods rely on sta-
ble isotope labeling while others use computational approaches to
obtain quantitative information from MS data (i.e. label-free quan-
tification). In principle, any method can be used for q-AP-MS as
long as it is accurate enough for unambiguous comparison of both
samples. Stable isotope-based approaches are usually more precise
than label-free approaches. For cell culture experiments, stable iso-
tope labeling by amino acids in cell culture (SILAC) is both the easiest
and most accurate stable isotope-based approach [26,27]. SILAC
can now even be used to label model organisms like flies and mice
[28,29]. Therefore, SILAC is generally the method of choice for q-
AP-MS experiments and also used in this protocol.
Label-free quantification has also been used recently for q-AP-
MS and is the method of choice when SILAC is not possible
[13,16,19,20]. However, care should be taken when selecting the
algorithm for label-free quantification. Although still popular,
spectral counting provides only a very rough estimate of protein
abundance and is thus error-prone . Label-free algorithms
based on pair-wise comparison of peptide peak intensities are
more accurate and therefore preferable .
2.3. What is a suitable control?
q-AP-MS relies on comparing the abundance of proteins co-
purified with the bait with a suitable control. This is a critical point
since any change in protein abundance between the actual AP-MS
sample and the control will be considered as a potential interaction
partner. The choice of a suitable control depends a lot on the con-
text of the experiment. In general, the closer the control resembles
the actual experiment the better. For example, if the goal is to
F.E. Paul et al./Methods 54 (2011) 387–395
identify proteins interacting with a modified peptide, the control
should be the same peptide in the non-modified form. In case of
endogenously expressed baits control cells should be transfected
with the empty expression vector. In the latter case, the same affin-
ity matrix must be used for both the experiment and the control.
This is important since different affinity matrixes have different
cross-reactivities. For example, different antibodies can cross-react
with different cellular proteins. Therefore, a ‘‘control’’ antibody,
Fig. 1. Experimentalworkflowforquantitativeinteractionproteomics.(A)Schemeforpull-downprocedurewithexogenousexpression ofthebaitprotein.Eitherlightorheavy
stable-isotope labeled cell lysate is incubated with tagged bait or control proteins, which has been crosslinked to the affinity matrix. After the pull-down beads are mixed and
captured bait/prey complexes are eluted and analyzed by LC-MS/MS. Crossover experiments are performed by swapping the lysate. (B) Scheme for affinity purification with
endogenous expression of the bait protein. Stable-isotope labeled cells are transiently transfected with an expression plasmid encoding a bait or control protein with an
appropriate tag. After bait expression, the tagged constructs are immunoprecipitated and the eluates are combined before subsequent MS sample preparation and LC-MS/MS
low heavy-to-light-ratios in the crossover experiments, whereas non-specific binders have 1:1 heavy-to-light ratio in both experimental conditions (A and B).
F.E. Paul et al./Methods 54 (2011) 387–395
even of the same isotype, is not a suitable control for an immuno-
2.4. Should crosslinking be used?
The results of q-AP-MS experiments should reflect the composi-
tion of endogenous protein complexes. This may, however, not be
true in all cases: Although q-AP-MS captures transient interactions,
very labile complexes may fall apart during purification. Con-
versely, PPIs may artificially form in a cell lysate between proteins
that do not interact in vivo. The latter point is also the reason why
even SILAC samples and their controls should generally be mixed
after affinity purification . Crosslinking offers the opportunity
to ‘‘freeze’’ PPIs inside cells before lysis. The most popular cross-
linker is formaldehyde: Due to its small size, formaldehyde can
permeate cell walls and membranes and induce efficient, revers-
ible cross-links between proteins . Crosslinking can also pro-
vide structural information about proteins and PPIs, although this
approach is challenging both technically and bioinformatically
. Due to these complications most q-AP-MS experiments are
performed without crosslinking of the proteins in the lysate.
A related question is whether or not the bait protein (for exog-
enous bait) or the antibody (for coIP) should be crosslinked cova-
lently to the affinity matrix. Without crosslinking, the bait
protein or antibody is often by far the most abundant protein in
the purified sample. Since mass spectrometry is mainly limited
by dynamic range, this may cause less abundant interaction part-
ners to escape detection. Thus, it is generally recommended to
crosslink the bait protein or the antibody to the affinity matrix.
For exogenous bait experiments we tested two alternatives: the
homobifunctional crosslinker dimethyl pimelimidate (DMP, )
and N-Hydroxysuccinimide (NHS) activated Sepharose. Both cross-
linkers clearly increased the number of identified interaction part-
ners, showing that coupling baits to the matrix is indeed beneficial
(see Fig. 2 and Section 4.). Many anti-tag-antibodies can be
Peptide and protein affinity tags commonly used in purification of fusion proteins.
Tag nameSequence OriginSize Binding partnerCommentReference
KRRWKKNFIAVSAANRFKKISSSGAL Rabbit skeletal muscle26 aa –
Calmodulin Binding in
EQKLISEEDLHuman oncogene c-
10 aa –
8 aa –
9 aa –
6 aa –
Anti-c-myc antibody (9E10)
FLAGDYKDDDDK Anti-FLAG antibody (M1,
Anti-HA antibody (12CA5)
HAYPYDVPDYA Hemaglutinin from
human influenza virus
Strep-tag II WSHPQFEK
8 aa –
K. pneumonia/E. coli
72/15 aa Biotinylation by birA (E. coli)
binding by Streptavidin/
169 aa Biotin Protein-ligase OR
Variation of the
kD = 1.2 nm 
238 aa –
40 kDaAmylose 
Rabbit + Staphylococcus
?20 kDa 1st step: IgG matrix 2nd
ProteinA + TEV
cleavage site + CBP
Fig. 2. Pulldown of Cdc42 effector proteins is most efficient with NHS-Sepharose.
The pull-downs of recombinantly expressed and nucleotide-loaded Cdc42-GTPcS
and Cdc42-GDP were carried out in a q-AP-MS experiment. Three experiments were
performed, first with glutathione beads without crosslinker, second with glutathi-
one beads with DMP crosslinking and third with NHS-activated Sepharose. Effector
proteins were pulled-down from SILAC labeled HeLa cell lysate. Depicted is the
number of proteins known to interact with Cdc42 in the list of thirty identified
proteins with the highest ratio. The highest amount of known effectors is recovered
using the NHS-Sepharose approach.
F.E. Paul et al./Methods 54 (2011) 387–395
purchased directly in a matrix-bound form. If crosslinked antibod-
ies are not available, it is usually advisable to crosslink them to
protein A or protein G Sepharose using DMP or another bifunc-
tional crosslinker (see  for a detailed overview of different
2.5. Which kind of biological replicates are advisory?
As for all biological experiments, replicates also increase confi-
dence in q-AP-MS experiments. In case of SILAC-based quantifica-
tion such a replicate is most conveniently performed as a so-called
crossover experiment with swapped isotope labels (Fig. 1). Swap-
ping the labels should result in reciprocal SILAC ratios for true
interaction partners (see Fig. 3 and Conclusions). In contrast, exog-
enous contaminants such as proteins derived from fetal bovine ser-
um are always unlabeled (i.e. light) and can thus easily be
identified. For pull-down experiments with exogenous baits a
crossover experiment can also control for differences in protein
abundance in the cell lysates used for the experiment. Such differ-
ences in cellular protein abundance could otherwise be misinter-
preted as specific interactions. Since label-free quantification is
generally less accurate than SILAC replicates are even more impor-
tant in this case. Label-free experiments should usually be per-
formed at least in triplicates since this permits statistical tests
like the Student’s t-test for reliable identification of interaction
3. Experimental protocols
3.1. General remarks
In this chapter, we describe a step-by-step procedure for the
identification of PPIs using either a pull-down with recombinant
bait protein or a coIP with a transfected bait protein. We use either
stable isotope-labeled HeLa or HEK293T cells but in principle other
cell lines that are amenable to SILAC can be utilized as well. We
also propose different cell lysis buffers (Section 3.2.2) for each
purification protocol, but again, other lysis buffers with protease
and phosphatase inhibitors can be used as well, if the conditions
are not too denaturing and work efficiently for the protein of inter-
est. A considerable practical advantage of q-AP-MS over tandem
affinity purification is that less material is needed. We routinely
use the protocols described here to identify interaction partners
from about 2 ? 107cells using in-solution digests. However, if
expression levels of bait protein in transient transfections are very
low or if known interaction partners are missed the experiments
can be scaled up by using more cells or bait protein. Another lim-
iting factor is the dynamic range of protein concentrations in the
precipitates. Therefore, a prefractionation step by SDS–PAGE can
enhance identification rates. Since only a fraction of the loaded
protein amount is recovered during in-gel digestion  it is often
advisable to prefractionate and scale up at the same time. In our
experience, both protocols can be scaled up easily to five to ten
15-cm plates per condition.
3.2.1. Cell Culture and stable isotope labeling by amino acids in cell
Medium: Dulbecco’s Modified Eagle’s Medium (DMEM) lacking
arginine and lysine (custom preparation from Gibco)
? Amino acids:
bridge Isotope Laboratories).12C614N4L-arginine and12C614N2
? Supplements: dialyzed fetal bovine serum (dFBS, Gibco), L-gluta-
mine (Gibco), penicillin/streptomycin (Invitrogen),
? Cells: adherent HeLa epithelial adenocarcinoma (LGC Promo-
chem) or adherent HEK293TN human embryonic kidney (Sys-
tem Biosciences) cells.
3.2.2. Cell lysis, pull-down, transfection, affinity purification and mass
? Cell lysis:
– Pull-down lysis buffer: 50 mM HEPES pH 7.5, 150 mM NaCl,
0.2% NP-40, Complete Protease Inhibitor cocktail (Roche)
1:25, Phosphatase Inhibitor Mix 1 (Sigma) 1:100
– Radioimmunoprecipitation (RIPA) buffer: 50 mM Tris–HCl
pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton-X 100, 1% Na-
Deoxycholate, 0.1% SDS, Benzonase (Sigma), Complete Pro-
tease Inhibitor cocktail (Roche) 1:25, Phosphatase Inhibitor
Mix 1 (Sigma) 1:100
– Dulbecco’s phosphate buffered saline (D-PBS, Invitrogen)
without calcium chloride and magnesium chloride
? Pull-down experiment
- Beads: NHS-activated Sepharose 4 Fast Flow (GE Healthcare)
- Chemicals: hydrochloric acid (Roth)
- Coupling buffer: the composition of the coupling buffer
depends on the ligand to be coupled. Phosphate or carbonate
based buffers with pH 6–9 are standard. Avoid primary
amines like Tris.
- Quenching buffer: 0.2 M ethanolamine pH 8.0.
- Low pH buffer (e.g. 0.1 M acetate, 0.5 M NaCl pH 5)
- High pH buffer (e.g. 0.1 M Tris HCl pH 8)
- Stringency wash buffer: e.g. 50 mM HEPES pH 7.5, 300 mM
NaCl, 0.2% NP-40 (low stringency)
- Pre-elution buffer: e.g. 3 mM HEPES pH 7.5, 300 mM NaCl
(reduces the buffer capacity to facilitate subsequent acidic
- Elution buffer: 0.1 M glycine pH 3.0
? Transient transfection of mammalian cells: linear polyethyleni-
mine (PEI ‘Max’, nominally Mw 40,000, Polysciences), 1 lg/ll in
dH2O stock solution
? Affinity purification: lMACS Epitope Isolation Kit and respec-
tive M columns (Miltenyi Biotec), elution buffer: 0.1 M glycine
? Protein precipitation: 2.5 M sodium acetate pH 5.0 (Merck),
Glycoblue (Ambion), 1 M Tris pH 8.2 (Roth), 100% LC-MS grade
? In-solution digest and LC–MS/MS analysis
– In-solution digest: Lysyl endopeptidase (Lys-C) (Wako) and
sequencing grade modified trypsin (Promega). Stop and go
extraction tips containing C18Empore disks (3 M)
– LC–MS/MS analysis: ReproSil-Pur C18-AQ 3 lm resin (Dr.
Maisch), LC–MS grade acetonitrile (Sigma), LC-MS grade
water (Sigma), LC-MS grade formic acid (Fluka).
3.3. Cell culture
HeLa or HEK293T cells are cultivated at 37 ?C with 5% CO2and
split every second or third day. SILAC media is essentially prepared
as described . Briefly, DMEM lacking arginine and lysine is sup-
plemented with 10% dialyzed fetal bovine serum, 4 mM glutamine
and 1% penicillin/streptomycin. To prepare heavy (H) SILAC media
we add 28 mg/l13C615N4L-arginine plus 49 mg/l13C615N2L-lysine.
Light (L) SILAC medium is prepared by adding the corresponding
non-labeled amino acids.
F.E. Paul et al./Methods 54 (2011) 387–395
HeLa or HEK293T cells are labeled in the corresponding SILAC
medium for at least seven cell divisions. For one quantitative
immunoprecipitation, usually one 15 cm dish of light and heavy
cells, respectively, is sufficient (2 ? 107cells of each SILAC state).
3.4. Exogenous expression of bait protein followed by pull-down
N-Hydroxysuccinimides (NHS) are activated esters which react
with amine groups to amides. The following protocol is based on
the standard protocol for NHS-activated Sepharose 4 Fast Flow
(GE Healthcare). The recombinant protein should not be stored in
solutions containing primary amines (e.g. Tris- or ammonium-
based buffers) because they will react with NHS and decrease
available binding sites. The protocol describes a single pull-down,
with a respective amount of 100–150 lg of recombinant protein
and 100 ll NHS-resin. The NHS protocol can also be performed
5–10 lg of antibody. A complete quantitative q-AP-MS experiment
30 ll NHS-Sepharoseand
requires 4 pull-downs (2 for the actual experiment and 2 for the
crossover experiment). Optimal ratios of slurry to antibody to
lysate are dependent on the protein used and should be experi-
Notes of volumes refer to the initial amount of slurry (e.g. ‘4 vol-
umes’ for an initial volume of 30 ll slurry means 120 ll). To avoid
hydrolysis of the activated ester all steps preceding coupling
should be performed without delays.
3.4.1. Preparation of NHS-beads
Pipette 100 ll of NHS-beads slurry into a 1.5 ml Eppendorf tube.
Centrifuge (1000g, 2 min) and remove the supernatant. To remove
isopropanol used for storing the beads wash once with 10–15 vol
cold 1 mM hydrochloric acid. Centrifuge again (1000g, 2 min, 4 ?C).
Wash twice with at least 4 volumes coupling buffer, centrifuge
(1000g, 2 min, 4 ?C). Check the pH and adjust if necessary with
(contaminant from serum)
exogenous expression of bait protein
followed by pull-down
endogenous expression of bait protein
followed by purification
700 697698 699
14-3-3 protein epsilon
14-3-3 protein epsilon
Hemoglobin subunit alpha
Fig. 3. Exemplary mass spectra of identified peptides. SILAC pairs for the straight and the corresponding crossover experiment are depicted next to each other. All peptides
contain one lysine, resulting in a mass difference of 8 Da between the light (white circle) and the corresponding heavy (black circle) peak. (A) Cdc42-GTPcS used as bait in a
pull-down specifically interacts with Synaptojanin, SILAC ratios swap according to the label state swap. (B) Cdc42-GDP used as bait in a pull-down specifically interacts with
Rho-GDI, SILAC ratios swap according to the label state swap. (C) The 14-3-3 protein shows a 1:1 ratio in the Cdc42 experiment, indicating non-specific binding. (D) Ataxin-1
used as the bait in an immunoprecipitation with swapped ratio in the crossover experiment. (E) The known Ataxin-1 binding protein 14-3-3 exhibits a ratio similar to the bait
protein itself, indicating a specific interaction with Ataxin-1. (F) The unlabeled serum protein hemoglobin shows very low heavy-to-light ratios in both experimental
conditions, indicating an non-specific contamination derived from serum.
F.E. Paul et al./Methods 54 (2011) 387–395
additional washing steps. The pH should not be acidic to avoid
denaturation of the recombinant protein. The pH can range be-
tween 6–9, higher pH will result in hydrolysis of the ester bound.
Mix the recombinant protein (or antibody) with coupling buffer
to a final volume of 0.5 vol and add the solution to the slurry. Beads
should always stay wet and be able to freely move within the dis-
persion. If beads tend to dry add coupling buffer (optimal ratio cou-
pling solution:medium is 0.5:1). Incubate by head over tail rotation
for 120 min at room temperature.
Centrifuge and remove the supernatant. Take a sample of the
supernatant and check for binding efficiency via western blot.
Add 2.5 vol quenching buffer. Incubate for 30 min at room temper-
ature by head over tail rotation. Quenching is needed to saturate
free binding sites.
3.4.5. Wash cycles
Centrifuge and remove the supernatant. Wash with alternating
wash buffers to remove not covalently coupled bait protein. To use
a low and a high pH buffer has proven to be practical. Repeat this
step three times.
Finally wash twice with at least 4 vol lysis buffer. The affinity
matrix is ready for use after this step. For storage add ethanol
(20%) or azide and store at 4 ?C. However, some coupled proteins
are not very stable, so immediate usage is normally advisable.
3.4.6. Pull-down from cell lysate
Save a sample of cell lysate (sample ‘‘Input’’). Add the cell lysate
to the Protein-Sepharose. Again, beads should stay wet and be able
to freely move within the dispersion (use a matrix:liquid phase ra-
tio of ?1:1, if beads seem too dry add a bit of lysis buffer). Incubate
the pull-down on a spinning wheel at 4 ?C (incubation time from
30 min to overnight, depending on the bait to prey affinity and sta-
bility). Centrifuge and save a sample of the supernatant (sample
3.4.7. Stringency wash
Pool the respective samples (two pull-downs from the
‘‘straight’’ experiment and the two pull-downs from the crossover).
Wash one to three times with 4 vol washing buffer (e.g. 50 mM
HEPES pH 7.5, 300 mM NaCl, 0.2% NP-40). If pull-downs result in
a high number of non-specific background binders (ribosomal pro-
teins, histones) increase the stringency by using a higher salt or
detergent concentration (0.5 M NaCl, 0.1% SDS).
3.4.8. Pre-elution wash
Wash once with at least 4 volumes of a low buffer non-
detergent solution. This step facilitates acidic elution of the bound
interaction partners for in-solution digestion. In case in-gel diges-
tion is used beads can be directly boiled in SDS–PAGE sample
buffer after the stringency wash.
Incubate the beads with 100 ll elution buffer, then incubate for
2 min on a spinning wheel or thermo shaker, check the pH (use a
pH indicator stick, the pH should be acidic, if not add more elution
buffer, incubate again). Centrifuge and save the supernatant, it
should contain the protein of interest and interaction partners.
Elute three times in total. Pool the three eluates and continue
immediately with protein precipitation.
3.4.10. Protein precipitation
Transfer the sample (volume not more than 300 ll) to a 2 ml
dust-free tube. Add 70 ll sodium acetate and 2 ll Glycoblue. The
addition of Glycoblue is recommended if the protein concentration
is low. Fill up to 2 ml with 100% ethanol and mix briefly by invert-
ing the Eppendorf tube. Incubate the precipitation overnight at
room temperature or at 4 ?C and centrifuge the next day with
20,000g at 4 ?C for at least 30 min. Remove the supernatant care-
fully and air dry the protein pellet. Continue with MS sample prep-
aration (Section 3.6).
3.4.11. Control pull-down efficiency
To check if the pull-down was efficient, compare the ‘‘input’’,
‘‘output’’ and bead sample by SDS–PAGE and western blotting.
Known interaction partners of the bait protein should be detected
in the input sample but significantly depleted in the output. A high
amount of interaction partners in the bead sample suggests ineffi-
3.5. Transient transfection of bait protein followed by
Here, we provide a protocol optimized for immunoprecipita-
tions of epitope-tagged proteins overexpressed in the HEK293T
system by using the lMACS Epitope Isolation Kit (Miltenyi Biotec).
These magnetic microbeads gave best results for our immunopre-
cipitations and offer other important advantages . However,
other affinity matrices may work as well. We use linear PEI as
the transfection reagent for HEK293T, as it is a highly efficient
(usually P 95%) and much cheaper than many other commercially
available alternatives. Different cell lines may require different
3.5.1. Transient transfection
Mix the expression plasmid with polyethylenimine in a 1:2
DNA to PEI ratio (transfection ratio may need optimization
depending on the cell line). For a 15 cm cell culture dish mix
15 lg plasmid DNA with 30 lg PEI and add 2 ml of serum-free
DMEM. Incubate for 30–45 min at room temperature to assure effi-
cient binding of the DNA to PEI. Add additional 2 ml of serum-free
DMEM and mix briefly before transferring your transfection setup
to the cells. Cell confluency should be 60–70% on the day of trans-
fection. For control transfections use an empty vector carrying only
the tag sequence or a more appropriate control (see Section 2.3).
3.5.2. Cell harvesting
Harvest the cells 24 h post-transfection. Aspire the medium,
wash once with ice-cold DPBS, then add 500 ll of ice-cold RIPA ly-
sis buffer. Use cell scrapers to harvest the cells and incubate them
for 30 min on ice. Cell debris should be cleared by centrifugation at
20,000g for 10 min at 4 ?C.
3.5.3. Immunoprecipitation from cell lysate
Add 50 ll oflMACS beads (directed against the tag of choice) to
each lysate from different SILAC conditions separately and incu-
bate on ice for 45 min.
3.5.4. Column preparation
Equilibrate the MACS column with 200 ll ice-cold RIPA buffer.
Transfer the IP slurries to the MACS column, the different IP setups
are now combined on the column.
Wash the lMACS column three times with 200 ll lysis buffer
and once with 100 ll Miltenyi wash buffer 2 (included in the Epi-
F.E. Paul et al./Methods 54 (2011) 387–395
tope Isolation kit), to reduce the salt concentration before elution
of the proteins.
The proteins are eluted from the lMACS column in a dust-free
2 ml tube by adding 300 ll elution buffer to the column. Continue
with protein precipitation as described in Section 3.4.10. Precipi-
tated proteins are ready for MS sample preparation (Section 3.6).
3.6. In-solution digest
MS sample preparation including reduction and alkylation of
cysteine residues is done as described . Lysyl endopeptidase
(Lys-C) (Wako) and sequencing grade modified trypsin (Promega)
are used for in-solution digestion. Stop and go extraction tips con-
taining C18Empore disks (3 M) are used to purify and store peptide
On-line LC-MS/MS analysis is performed as described previ-
ously . In brief, peptide mixtures are separated by reversed
phase chromatography using the Eksigent NanoLC-1D Plus system
(Eksigent) on in-house manufactured 10-cm fritless silica micro-
columns with an inner diameter of 75 lm. Columns are packed
with ReproSil-Pur C18-AQ 3 lm resin (Dr. Maisch GmbH) . Sep-
aration is performed using a 10–60% ACN gradient (240 min) with
0.5% formic acid at a flow rate of 200 nl/min. Eluting peptides are
directly ionized by electrospray ionization and transferred into
the orifice of a linear trap quadrupole Orbitrap hybrid mass spec-
trometer (Thermo Fisher Scientific). Mass spectrometry is per-
formed in the data-dependent mode with one full scan in the
Orbitrap (m/z 300–1700; resolution of 60,000; AGC target value
of 1 ? 106). The five most intense ions with a charge state greater
than 1 are selected (target value 3000; monoisotopic precursor
selection enabled) and fragmented in the linear trap quadrupole
using CID (35% normalized collision energy and wideband activa-
tion enabled). Dynamic exclusion for selected precursor ions is
3.8. Processing of MS data
The MaxQuant software package (version 184.108.40.206) is used to
identify and quantify proteins [38,39]. SILAC duplets are extracted
from isotope patterns, recalibrated, and quantified by the Quant
module (heavy label Arg10 and Lys8, maximum of three labeled
amino acids per peptide, polymer detection enabled, and top six
MS/MS peaks per 100 Da). Peak lists are searched on a Mascot
search engine (version 2.2, Matrix Science) against an in-house cu-
rated database of H. sapiens (IPI.Human, version 3.64) plus com-
mon contaminants. When performing interaction studies with
recombinant proteins expressed in a different organism (e.g.
E. coli) the database should also contain all protein entries from
this organism. This is important since the purified recombinant
protein often still contains trace amount of contaminants derived
from this species, and these may be falsely assigned to for example
human proteins (when searching a human database). All protein
sequences are also reversed to generate a target-decoy database
. Carbamidomethylation of cysteine is selected as a fixed mod-
ification, and oxidation of methionine and acetylation of the pro-
tein N-terminus are used as variable modifications. Lys-C and
trypsin are selected as proteases (full specificity) with a maximum
of two missed cleavages. A mass tolerance of 0.5 Da is selected for
fragment ions. A minimum of six amino acids per identified pep-
tide and at least one peptide per protein group are required. The
false discovery rate is set to 1% at both the peptide and protein
levels. Protein ratios are calculated from the median of all normal-
ized peptide ratios using only unique peptides or peptides assigned
to the protein group with the highest number of peptides. Only
protein groups with at least three SILAC counts are considered
for further analysis.
Combining classical pull-down or coIP techniques with SILAC-
based quantitative proteomics is a powerful approach to detect
PPIs with high sensitivity and, perhaps more importantly, high
specificity. As an example for exogenous bait expression, we show
that recombinantly expressed GTPcS-loaded Cdc42 versus the
GDP-loaded form can be used to identify specific interaction part-
ners of both forms of this Rho-GTPase (Fig. 3A–C). As an example
for cellular expression of tagged bait we used Ataxin-1 and show
specific interaction with 14-3-3 protein epsilon (Fig. 3D–F). In all
cases, heavy-to-light SILAC ratios of specific interaction partners
are inverted in the crossover experiment, validating their specific
binding. In contrast, non-specific background binders show a 1:1
ratio in both the straight and the crossover experiment (Fig. 3C).
A contaminating serum protein is always detected in the light
(i.e. unlabeled) form and can thus easily be excluded (Fig. 3F).
We also show that covalently crosslinking the bait to the matrix
yields a higher number of literature-reported interaction partners
(Fig. 2). We assume that crosslinking reduces the amount of bait-
derived peptides in the sample which facilitates MS analysis. The
highest number of known interaction partners was detected using
the NHS-approach, where 10 interaction partners were exclusively
found. Only two interaction partners were not covered by the NHS-
based pull-down. In summary, the two protocols described and the
background information provided should facilitate adapting the
generic q-AP-MS strategy for a wide range of biological questions.
 J.D. Scott, T. Pawson, Science 326 (2009) 1220–1224.
 U. de Lichtenberg, L.J. Jensen, S. Brunak, P. Bork, Science 307 (2005) 724–727.
 S. Fields, O. Song, Nature 340 (1989) 245–246.
 U. Stelzl, E.E. Wanker, Curr. Opin. Chem. Biol. 10 (2006) 551–558.
 Schlessinger, J., Lemmon, M.A. (2003). Sci. STKE 2003, RE12.
 A.C. Gingras, M. Gstaiger, B. Raught, R. Aebersold, Nat. Rev. Mol. Cell Biol. 8
 R. Aebersold, M. Mann, Nature 422 (2003) 198–207.
 X. Han, A. Aslanian, J.R. Yates 3rd, Curr. Opin. Chem. Biol. 12 (2008) 483–490.
 G. Rigaut, A. Shevchenko, B. Rutz, M. Wilm, M. Mann, B. Seraphin, Nat.
Biotechnol. 17 (1999) 1030–1032.
 M. Selbach, F.E. Paul, S. Brandt, P. Guye, O. Daumke, S. Backert, C. Dehio, M.
Mann, Cell Host Microbe 5 (2009) 397–403.
 B. Blagoev, I. Kratchmarova, S.E. Ong, M. Nielsen, L.J. Foster, M. Mann, Nat.
Biotechnol. 21 (2003) 315–318.
 M. Vermeulen, H.C. Eberl, F. Matarese, H. Marks, S. Denissov, F. Butter, K.K. Lee,
J.V. Olsen, A.A. Hyman, H.G. Stunnenberg, M. Mann, Cell 142 (2010) 967–980.
 N.C. Hubner, A.W. Bird, J. Cox, B. Splettstoesser, P. Bandilla, I. Poser, A. Hyman,
M. Mann, J. Cell Biol. 189 (2009) 739–754.
 M. Selbach, M. Mann, Nat. Methods 3 (2006) 981–983.
 M. Bantscheff, D. Eberhard, Y. Abraham, S. Bastuck, M. Boesche, S. Hobson, T.
Mathieson, J. Perrin, M. Raida, C. Rau, V. Reader, G. Sweetman, A. Bauer, T.
Bouwmeester, C. Hopf, U. Kruse, G. Neubauer, N. Ramsden, J. Rick, B. Kuster, G.
Drewes, Nat. Biotechnol. 25 (2007) 1035–1044.
 M.E. Sowa, E.J. Bennett, S.P. Gygi, J.W. Harper, Cell 138 (2009) 389–403.
 F. Mousson, A. Kolkman, W.W. Pijnappel, H.T. Timmers, A.J. Heck, Mol. Cell.
Proteomics 7 (2008) 845–852.
 T. Wang, S. Gu, T. Ronni, Y.C. Du, X. Chen, J. Proteome Res. 4 (2005) 941–949.
 A.C. Paoletti, T.J. Parmely, C. Tomomori-Sato, S. Sato, D. Zhu, R.C. Conaway, J.W.
Conaway, L. Florens, M.P. Washburn, Proc. Natl Acad. Sci. USA 103 (2006)
 O. Rinner, L.N. Mueller, M. Hubalek, M. Muller, M. Gstaiger, R. Aebersold, Nat.
Biotechnol. 25 (2007) 345–352.
 M. Vermeulen, N.C. Hubner, M. Mann, Curr. Opin. Biotechnol. 19 (2008) 331–
 W.X. Schulze, M. Mann, J. Biol. Chem. 279 (2004) 10756–10764.
 K. Terpe, Appl. Microbiol. Biotechnol. 60 (2003) 523–533.
 S.E. Ong, M. Mann, Nat. Chem. Biol. 1 (2005) 252–262.
 M. Wilm, Proteomics 9 (2009) 4590–4605.
F.E. Paul et al./Methods 54 (2011) 387–395
 S.E. Ong, M. Mann, Nat. Protoc. 1 (2006) 2650–2660. Download full-text
 M. Mann, Nat. Rev. Mol. Cell Biol. 7 (2006) 952–958.
 M.D. Sury, J.X. Chen, M. Selbach, Mol. Cell. Proteomics 9 (2010) 2173–2183.
 M. Kruger, M. Moser, S. Ussar, I. Thievessen, C.A. Luber, F. Forner, S. Schmidt, S.
Zanivan, R. Fassler, M. Mann, Cell 134 (2008) 353–364.
 G. Liu, J. Zhang, B. Larsen, C. Stark, A. Breitkreutz, Z.Y. Lin, B.J. Breitkreutz, Y.
Ding, K. Colwill, A. Pasculescu, T. Pawson, J.L. Wrana, A.I. Nesvizhskii, B.
Raught, M. Tyers, A.C. Gingras, Nat. Biotechnol. 28 (2010) 1015–1017.
 B.W. Sutherland, J. Toews, J. Kast, J. Mass Spectrom. 43 (2008) 699–715.
 Rappsilber, J. (2010). J. Struct. Biol. in press.
 T.H. Sisson, C.W. Castor, J. Immunol. Methods 127 (1990) 215–220.
 G. Mattson, E. Conklin, S. Desai, G. Nielander, M.D. Savage, S. Morgensen, Mol.
Biol. Rep. 17 (1993) 167–183.
 A. Shevchenko, H. Tomas, J. Havlis, J.V. Olsen, M. Mann, Nat. Protoc. 1 (2006)
 J. Rappsilber, Y. Ishihama, M. Mann, Anal. Chem. 75 (2003) 663–670.
 Y. Ishihama, J. Rappsilber, J.S. Andersen, M. Mann, J. Chromatogr. A 979 (2002)
 J. Cox, M. Mann, Nat. Biotechnol. 26 (2008) 1367–1372.
 J. Cox, I. Matic, M. Hilger, N. Nagaraj, M. Selbach, J.V. Olsen, M. Mann, Nat.
Protoc. 4 (2009) 698–705.
 J.E. Elias, S.P. Gygi, Nat. Methods 4 (2007) 207–214.
 R.E. Stofko-Hahn, D.W. Carr, J.D. Scott, FEBS Lett. 302 (1992) 274–278.
 G.I. Evan, G.K. Lewis, G. Ramsay, J.M. Bishop, Mol. Cell. Biol. 5 (1985) 3610–
 T.P. Hopp, Bio/Technology 6 (1988) 1204–1210.
 U.K. Pati, Gene 114 (1992) 285–288.
 E. Hochuli, H. Dobeli, A. Schacher, J. Chromatogr. 411 (1987) 177–184.
 B. Zakeri, M. Howarth, J. Am. Chem. Soc. 132 (2010) 4526–4527.
 T.G. Schmidt, A. Skerra, Protein Eng. 6 (1993) 109–122.
 A. Tirat, F. Freuler, T. Stettler, L.M. Mayr, L. Leder, Int. J. Biol. Macromol. 39
 S. Sueda, H. Tanaka, M. Yamagishi, Anal. Biochem. 393 (2009) 189–195.
 F.G. Prendergast, K.G. Mann, Biochemistry 17 (1978) 3448–3453.
 M. Chalfie, Y. Tu, G. Euskirchen, W.W. Ward, D.C. Prasher, Science 263 (1994)
 D.B. Smith, K.S. Johnson, Gene 67 (1988) 31–40.
 M.G. Taylor, H. Bushara, A. Capron, S. Xu, A. Butterworth, in: S.H.P. Jepsen, R.
Klein, D. Taylor (Eds.), Science and technology for development-health-second
programme (1987–1991). Summaries of the final reports of the research
contracts, Parasitology European Commission DGXII, 1994, pp. A:180–A:200.
 P. Duplay, H. Bedouelle, A. Fowler, I. Zabin, W. Saurin, M. Hofnung, J. Biol.
Chem. 259 (1988) 10606–11061.
F.E. Paul et al./Methods 54 (2011) 387–395