Heterogeneity and dynamics in the assembly of the
Heat Shock Protein 90 chaperone complexes
Ima-obong Ebonga,1, Nina Morgnera,1, Min Zhoua, Marco A. Saraivab, Soumya Daturpallib,
Sophie E. Jacksonb,2, and Carol V. Robinsona,2
aDepartment of Chemistry, University of Oxford, South Parks Road, Oxford OX1 3TA, United Kingdom; and
Cambridge, Lensfield Road, Cambridge CB2 1EW, United Kingdom
bDepartment of Chemistry, University of
Edited by F. Ulrich Hartl, Max Planck Institute of Biochemistry, Martinsried, Germany, and approved September 12, 2011 (received for review April 19, 2011)
The Hsp90 cycle depends on the coordinated activity of a range
of cochaperones, including Hop, Hsp70 and peptidyl-prolyl iso-
merases such as FKBP52. Using mass spectrometry, we investigate
the order of addition of these cochaperones and their effects on the
stoichiometry and composition of the resulting Hsp90-containing
complexes. Our results show that monomeric Hop binds specifically
to the Hsp90 dimer whereas FKBP52 binds to both monomeric and
dimeric forms of Hsp90. By preforming Hsp90 complexes with either
Hop, followed by addition of FKBP52, or with FKBP52 and subse-
quent addition of Hop, we monitor the formation of a predominant
asymmetric ternary complex containing both cochaperones. This
asymmetric complex is subsequently able to interact with the
chaperone Hsp70 to form quaternary complexes containing all
four proteins. Monitoring the population of these complexes during
their formation and at equilibrium allows us to model the complex
formation and to extract 14 different KDvalues. This simultaneous
calculation of the KDs from a complex system with the same meth-
od, from eight deferent datasets under the same buffer conditions
delivers a self-consistent set of values. In this case, the KDvalues
and provide a rationale for the cellular heterogeneity and preva-
lence of intermediates in the Hsp90 chaperone cycle.
dissociation constants ∣ interaction network
such as kinases and steroid hormone receptors (1–3). The activa-
tion of Hsp90 clients relies on interaction with its cochaperones
which facilitate different stages of the cycle (4–6). In human cells,
Hsp90 cochaperones include Hsp70, Hop, Aha1, Cdc37, p23, and
peptidyl-propyl isomerases (PPIases) such as FKBP51, FKBP52,
and CyP40. The exact complexes that are formed between Hsp90
and cochaperones depend on the client protein. Cdc37 appears to
be specific towards kinase clients whilst p23 is associated with the
binding and activation of steroid hormone receptors. Despite a
wealth of research, the exact complexes formed during the cycle
are not well established and their stoichiometry remains contro-
versial (6, 7).
Hsp90 is a homodimer containing three structural domains.
The N-terminal domain contains the ATP-binding site and binds
cochaperones including p23. The middle domain is suggested as
the client protein binding region. The C-terminal domain encom-
passes the dimerization interface and also contains a C-terminal
MEEVD motif which binds tetratricopeptide repeat (TPR)-
containing cochaperones (8). The ATPase activity of Hsp90 is
essential for its function and ATP hydrolysis is linked with a con-
formational rearrangement in which the homodimer changes
from an open conformation to a closed state in which the two
N-terminal domains interact (5, 6, 9–12). A general model was
proposed for the activation of steroid hormone receptors (SHR)
by the Hsp90 chaperone machinery (13, 14) in which the SHR
binds to the Hsp40/Hsp70 chaperones and is then transferred to
Hsp90 through the action of Hop which binds simultaneously
to Hsp70 and Hsp90 through its different TPR domains (15).
sp90 selectively interacts with and regulates the activation of
many client proteins involved in cellular signalling pathways
Subsequent binding of PPIases such as FKBP52 and cochaperone
p23 leads to a mature complex in which the client protein is cap-
able of being activated by binding of ligands and cofactors (13).
The Hsp90 dimer has two C-terminal MEEVD motifs which
are the primary binding sites for TPR-containing cochaperones
such as Hop and FKBP52 (16, 17). Hop has been established
as a crucial component of the Hsp90 cycle forming a link that
facilitates client protein transfer from Hsp70 to Hsp90 (15, 18).
Hop binding is also known to inhibit the ATPase activity of Hsp90
by stabilizing the open conformation (6, 19). When bound to the
Hsp90 dimer however there is conflicting evidence showing Hop
bindsas a dimer (6, 7,20, 21)ora monomer (22). Some molecular
details of how the TPR domains in FKBP52 and Hop bind to
the C-terminal MEEVD motifs of Hsp90 are known from crystal
structures of isolated TPR domains in complex with short
MEEVD-containing peptides (17). However, many details of
their binding in the context of the full-length Hsp90 and cochaper-
ones remain poorly understood. In particular, as Hsp90 is a homo-
dimer containing two MEEVD motifs, little is known about the
ternary complexes that might form with the different TPR cocha-
perones. There is a general view that Hop/FKBP52 may compete
for available binding sites in the Hsp90 dimer (13, 16, 23, 24).
The lack of reliable dissociation constants means that it is often
difficult to rationalize competitive binding and to predict the com-
plexes that form in the cellular environment.
Mass spectrometry (MS) is making significant inroads as an
adjunct to established structural biology tools by defining the
stoichiometry and interactions of protein complexes (25). Parti-
cularly significant have been recent applications of nanoflow
electrospray to polydisperse molecular chaperone-substrate com-
plexes wherein a snapshot of all complexes present at a given time
can be obtained within a single mass spectrum (26). This MS
approach should therefore enable investigation of the dynamics
of a protein system complicated by the formation of multiple
species in solution. By correlating the intensity of the MS peaks,
assigned to proteins and their complexes, with their concentra-
tion in solution we show here that it is possible to calculate KD
values for coexisting species. In this study, we investigate the het-
erogeneity and dynamics of the Hsp90 complexes formed in the
presence of Hop, FKBP52, and Hsp70. Although exemplified
with part of the Hsp90 reaction cycle, the approach detailed here
is applicable to other protein systems and is particularly suited
Author contributions: M.A.S., S.E.J., and C.V.R. designed research; I.E., N.M., and M.Z.
performed research; M.A.S. and S.D. contributed new reagents/analytic tools; I.E. and
N.M. analyzed data; and I.E., N.M., M.Z., S.E.J., and C.V.R. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Freely available online through the PNAS open access option.
1I.E. and N.M. contributed equally to this work.
2To whom correspondence may be addressed. E-mail: firstname.lastname@example.org or carol.robinson@
This article contains supporting information online at www.pnas.org/lookup/suppl/
www.pnas.org/cgi/doi/10.1073/pnas.1106261108PNAS ∣ November 1, 2011 ∣ vol. 108 ∣ no. 44 ∣ 17939–17944
to assemblies that are complicated by heterogeneity and polydis-
By manipulating the ratio of Hsp90 and Hop, we find that the
predominant complex is the ðHsp90Þ2ðHopÞ1although up to two
molecules of Hop can bind in excess of Hop. By preforming com-
plexes of the type ðHsp90Þ2ðHopÞn¼1 or 2and adding FKBP52, we
observe how FKBP52 competes with Hop for available Hsp90
binding sites to form the asymmetric ternary complex ðHsp90Þ
2ðHopÞ1ðFKBP52Þ1. Similarly, this complex can be assembled by
addition of Hop to a preformed ðHsp90Þ2ðFKBP52Þncomplex.
By monitoring the formation of these complexes in real time, we
calculate KDvalues for all ten complexes using eight different
datasets. Additionally, in the presence of Hsp70 we identify com-
plexes that are competent to bind Hsp70. Overall by monitoring
these heterogeneous chaperone assemblies we reveal a self-consis-
tent set of KDvalues which together with their established cellular
concentrations, allows us to predict the multiple Hsp90 complexes
existing in a cellular environment.
Hsp90 Binds Hop with a 2∶1 or 2∶2 Stoichiometry. We first recorded
mass spectra of all four proteins individually (Hsp90, Hsp70,
Hop, and FKBP52) under nondenaturing MS conditions (27).
Hsp90 revealed well resolved charge state series corresponding
to monomeric and predominantly dimeric Hsp90, in line with
previous MS studies (28, 29) (SI Appendix, Fig. S1A). The MS of
Hop revealed a mass corresponding to a monomer and no higher
oligomers (SI Appendix, Fig. S1B). For Hop, FKBP52, and Hsp70
measured masses corresponded in each case to a monomer and
no higher oligomers were detected (SI Appendix, Fig. S1 B–D).
All calculated and experimental masses, with the exception of
Hsp90, confirmed these proteins are monomeric under the con-
ditions used here (SI Appendix, Table S1).
To determine the binding stoichiometry of Hop to Hsp90,
we varied the concentration of both components. First the Hop
concentration was kept at 1 μM and the concentration of the
Hsp90 dimer ðHsp90Þ2varied from 1 to 5 μM. At equimolar
concentrations of Hop and ðHsp90Þ2, the dominant complex
was ðHsp90Þ2ðHopÞ1as well as Hsp90 monomer ðHsp90Þ1and
unbound Hsp90 dimer ðHsp90Þ2(SI Appendix, Fig. S1E). No free
Hop was seen, implying that all Hop molecules were bound in
the complex. Higher Hsp90 concentrations triggered no changes
in stoichiometry, the ðHsp90Þ2ðHopÞ1remained the prevalent and
only complex formed. The reverse experiment was also performed
using a constant ðHsp90Þ2concentration of 1 μM and increasing
the Hop concentration from 1 to 5 μM. We found that with excess
Hop (2 μM) an additional complex of ðHsp90Þ2ðHopÞ2formed;
concomitantly the concentration of free Hsp90 was depleted (SI
Appendix, Fig. S1F). A further increase in Hop concentration re-
sulted in the ðHsp90Þ2ðHopÞ2complex becoming more dominant
than the ðHsp90Þ2ðHopÞ1complex. No other higher-order com-
plexes were observed.
Because the Hsp90 dimer adopts an open conformation in the
presence of ADP and converts to a closed state when ATP is
Addition of 100 μM ADP to the Hop and Hsp90 complexes
showed no changes in the stoichiometry or ratios ofthe complexes.
Given that the KDs for ATP and AMPpNp are reported to be
higher than for ADP (30, 31), we also examined binding at higher
nucleotide concentration (300 μM). At this nucleotide concentra-
tion and using a KDvalue of 148 μM (30) and an Hsp90 monomer
concentration of 2 μM we calculate that >65% of the Hsp90 bind-
ing sites will be occupied. While the peaks are significantly broad-
er, due to multiple nonspecific attachments of nucleotide, there
is no change in the ratio of the complexes formed (SI Appendix,
Fig. S2 A–F). We conclude that while both Hsp90 subunits within
the dimer can bind a Hop monomer, the stoichiometry and ratios
of the complexes formed is nucleotide independent.
Hsp90 Can Bind Simultaneously to FKBP52 and Hop. We established
the interactions of ðHsp90Þ2with FKBP52, by incubating equimo-
lar concentrationsofðHsp90Þ2and FKBP52. The spectra recorded
revealed peaks that corresponded to the ðHsp90Þ2ðFKBP52Þ1and
ðHsp90Þ2ðFKBP52Þ2complexes (SI Appendix, Fig. S3A). We also
observed minor peaks for both FKBP52 and Hsp90 monomers and
an unexpected complex of ðHsp90Þ1ðFKBP52Þ1. It is noteworthy
that the analogous complex ðHsp90Þ1ðHopÞ1was not observed
for Hop binding to Hsp90.
We then examined competitive binding of Hop and FKBP52
to the Hsp90 dimer by forming either the ðHsp90Þ2ðFKBP52Þnor
ðHsp90Þ2ðHopÞncomplexes respectively at different molar con-
complexes of ðHsp90Þ2ðFKBP52Þnat equimolar concentrations of
FKBP52andHsp90 dimer (both 1 μM)and subsequentlyadded an
equimolar amount of Hop (SI Appendix, Fig. S3A). In the second
experiment, a binary complex of ðHsp90Þ2ðHopÞ1, was preformed
at equimolar concentrations and challenged with an equimolar
amount of FKBP52 (SI Appendix, Fig. S3B). In the third set of
experiments, we formed complexes of ðHsp90Þ2ðHopÞ1and
ðHsp90Þ2ðHopÞ2with 1 μM Hsp90 dimer and a twofold excess of
Hop(2μM)andthenchallengedthese with1μMFKBP52 (Fig.1).
In all three experiments, spectra were recorded at regular time
intervals up to 1 h after addition of the competing protein.
In the first experiment, the asymmetric ternary complex
ðHsp90Þ2ðFKBP52Þ1ðHopÞ1was clearly visible after 5 min (SI
Appendix, Fig. S3A). Concomitantly with the formation of this
complex, the relative abundance of the ðHsp90Þ2ðHopÞ1complex
plexes. MS recorded after t ¼ 0 to t ¼ 45 min showing the formation of the
asymmetric complex containing both Hop and FKBP52. ðHsp90Þ2ðHopÞ1and
ðHsp90Þ2ðHopÞ2were formed with twofold excess of Hop prior to the addition
ofFKBP52.Chargestatescoloredasfollows:Hop (light green),Hsp90monomer
and dimer (gray and black respectively), ðHsp90Þ2ðHopÞ1
ðHsp90Þ2ðHopÞ2purple, ðHsp90Þ1ðFKBP52Þ1yellow, ðHsp90Þ2ðFKBP2Þ1blue,
and ðHsp90Þ2ðHopÞ1ðFKBP52Þ1red. Same colors used throughout.
Hsp90, Hop, and FKBP52 interact to form binary and ternary com-
www.pnas.org/cgi/doi/10.1073/pnas.1106261108Ebong et al.
was also dominant compared to the ðHsp90Þ2ðFKBP52Þ1. In the
second experiment, addition of equimolar FKBP52 to ðHsp90Þ2
ðHopÞncomplexes preformed at equimolar concentrations re-
vealed peaks corresponding ðHsp90Þ2ðFKBP52Þnand the asym-
the ðHsp90Þ1ðFKBP52Þ1complex. If we compare the abundance
of the Hsp90 dimer with only one binding partner, we find the
ðHsp90Þ2ðHopÞ1is more prevalent than ðHsp90Þ2ðFKBP52Þ1
complex throughout the time course of both experiments. In the
third experiment, addition of FKBP52 to complexes containing
ðHsp90Þ2ðHopÞ1and ðHsp90Þ2ðHopÞ2similarly leads to formation
of the asymmetric ternary complex, as well as the same complexes
observed in the first two experiments. In this case, the ðHsp90Þ2
ðHopÞ2complex persists throughout the time course showing
Hop was not displaced, but rather FKBP52 binds on to the already
formed ðHsp90Þ2ðHopÞ1complex (Fig. 1, SI Appendix, Fig. S4A).
To test stability of these complexes formed in the third experiment,
we also recorded spectra after 3 h (SI Appendix, Fig. S4B). The
same complexes prevailed allowing us to conclude that the asym-
metric complex forms readily from a variety of solution conditions
and is stable for at least 3 h. As a control, we also tested for an
interaction between FKBP52 and Hop alone. No binding was
observed validating this control experiment and our observation
of specific complexes for Hsp90 with Hop or Hsp90 with FKBP52.
To compare the proportion of complexes formed as a function
of the total protein components including individual proteins and
complexes, we employed a deconvolution algorithm (Massign)
(32, 33). An example is shown for a spectrum recorded 15 min
after ðHsp90Þ2ðHopÞncomplexes were challenged with an equi-
molar ratio of FKBP52 (Fig. 2A). After 15 min the asymmetric
ternary ðHsp90Þ2ðFKBP52Þ1ðHopÞ1 and ðHsp90Þ2ðFKBP52Þ1
complexes represent 19% and 16% of the total population,
respectively. Similarly, we calculated the relative abundance of
all other components and plotted their populations as a function
of time (Fig. 2B). We find the asymmetric ternary complex forms
rapidly after addition of FKBP52. We also find that the homo-
dimeric ðHsp90Þ2becomes depleted over timeimplying that more
of the Hsp90 dimer is forming complexes with the TPR-contain-
ing cochaperones. The concentration of the ðHsp90Þ2ðHopÞ1
complex is also depleted over time while the ðHsp90Þ2ðHopÞ2re-
mains constant throughout implying that Hop is not displaced.
The heterodimer complex ðHsp90Þ1ðFKBP52Þ1forms while Hop
is unable to form an ðHsp90Þ1ðHopÞ1complex in the same fash-
ion. Our data shows both TPR-containing proteins readily bind
to an available site on the Hsp90 dimer, despite prior binding of
another, to form the asymmetric ternary complex.
Hsp90-Hop Complexes Promote Subsequent Binding of Hsp70. The
presence of multiple coexisting complexes prompts investigation
of complexes that are competent to bind Hsp70. Initially, we in-
vestigated binary interactions between Hsp70 and Hop or Hsp70
and Hsp90. At equimolar or higher concentrations of Hop, no
interaction was seen between Hsp70 and Hop in the absence of
nucleotides. The presence of 100 μM ADP however induced
formation of a weak 1∶1 complex of the Hsp70∶Hop heterodimer
(SI Appendix, Fig. S5A). Subsequently, we also examined whether
there was any direct interaction between Hsp90 and Hsp70 but no
binding was observed (SI Appendix, Fig. S5B). In contrast, on
addition of Hsp90 to Hsp70 in the presence of equimolar amount
of Hop, we observed a species with a measured mass correspond-
ing to an intermediate complex containing ðHsp90Þ2ðHopÞ1
ðHsp70Þ1(Fig. 3A). Also present was a predominant complex of
and ðHsp90Þ2. Other minor complexes containing ðHsp90Þ2
ðHopÞ2ðHsp70Þ1and ðHsp90Þ2ðHopÞ2ðHsp70Þ2were also formed
(SI Appendix, Fig. S6). As observed in the binary experiments,
the presence of ADP revealed no changes in the ratios of the
complexes formed (SI Appendix, Fig. S5 C and D). These data
show that Hsp70 binding to Hop is facilitated by formation of the
ðHsp90Þ2ðHopÞncomplexes because Hop alone showed only a
weak interaction with Hsp70 in the presence of ADP and was
unable to bind directly to Hsp70 in the absence of nucleotide.
To examine if FKBP52 could bind to the intermediate com-
plex, we first formed complexes of ðHsp90Þ2ðHopÞnðHsp70Þnand
subsequently added FKBP52 (Fig. 3B). Results showed unambi-
guously an additional complex that corresponds to ðFKBP52Þ1
ðHsp90Þ2ðHopÞ1ðHsp70Þ1(SI Appendix, Fig. S7). The observation
of Hsp70-containing complexes of ðHsp90Þ2ðHopÞ2ðHsp70Þ1,
ðHsp70Þ1allows us to conclude that the binding of Hsp70 is
directly linked to the number of Hop subunits bound to ðHsp90Þ2.
recorded 15 min after addition of an equimolar ratio of FKBP52 to
ðHsp90Þ2ðHopÞn(black) is compared with a simulated spectrum (blue) formed
by summing the deconvoluted spectra of all components (SI Appendix,
Fig. S3). The composition, relative intensity (%), and mass of each component
are given. (B) Kinetic model established from the relative intensity of all
components plotted against time (SI Appendix). The simulated kinetics
(lines) are shown together with the experimental data (symbols) from 0 to
40 min. The data with dashed lines is t ¼ 15 min.
Monitoring the formation of Hsp90 complexes in real-time (A) MS
Ebong et al.PNAS
November 1, 2011
A Kinetic Model for Hsp90 Complexes. A model was constructed to
show all feasible reaction pathways (SI Appendix, Fig. S8) and
highlight all possible complexes observed in our experiments
(Fig. 4). Starting with the Hsp90 dimer at micromolar concentra-
tions (green oval) a population of monomeric Hsp90 is formed
which is able to interact directly with FKBP52 but not Hsp70 or
Hop.Dimeric Hsp90isabletobindeither FKBP52orHoptoform
binary complexes of ðHsp90Þ2ðFKBP52Þ1or ðHsp90Þ2ðHopÞ1.
Having formed binary complexes containing the Hsp90 dimer
with either FKBP52 or Hop both complexes can bind another
copy of the same cochaperone to form ðHsp90Þ2ðFKBP52Þ2or
ðHsp90Þ2ðHopÞ2or bind the other cochaperone to form ðHsp90Þ2
ðFKBP52Þ1ðHopÞ1. The latter is favored statistically given equal
protein concentrations of Hop and FKBP52. Our data shows a
maximum of two copies of Hop or FKBP52 can bind per Hsp90
dimer. Interestingly, FKBP52 cannot bind complexes containing
two Hop monomers and Hsp70 cannot bind to any complex that
does not contain Hop and Hsp90.
To establish which complexes likely form under cellular condi-
tions, we determined KDvalues using the intensity ratios of the
10 different species and monitored their population as a function
of time (Fig. 2). We first established that the intensities of the
peaks assigned to each Hsp90-containing complex are related
to their concentrations in solution using an internal standard over
appropriate concentration ranges (SI Appendix, Fig. S9). We
assume that association and dissociation events follow second-
and first-order kinetics respectively (SI Appendix). KDvalues are
calculated by satisfying equations relating to the concentration of
all components at every time point. We performed eight different
sets of experiments including different concentrations, different
order of addition as well as replicate experiments (SI Appendix,
Table S2). This data yields a series of KDvalues that are reprodu-
cible within error between the different experimental conditions
as well as replicate experiments (Fig. 4). All KDvalues that deter-
mine complex formations are in the range ∼10–250 nM.
These KDvalues reveal several interesting features. The bind-
ing of the first Hop to a binding site on the free ðHsp90Þ2dimer to
form ðHsp90Þ2ðHopÞ1has a KDof 15 nM, whilst binding of the
second Hop to form ðHsp90Þ2ðHopÞ2has a KDof 55 nM. Binding
of FKBP52 to the ðHsp90Þ2to form either ðHsp90Þ2ðFKBP52Þ1
or ðHsp90Þ2ðFKBP52Þ2is slightly less favorable than for Hop
with corresponding KDvaluesof 145 nM and 120 nM, respectively.
Comparing the value obtained for Hop binding to ðHsp90Þ2
ðFKBP52Þ1(KD¼ 15 nM), with that obtained for Hop binding
to free ðHsp90Þ2(KD¼ 15 nM) we deduce that prior binding
of FKBP52 has no effect on subsequent binding of Hop to Hsp90.
Likewise, adding FKBP52 to ðHsp90Þ2ðHopÞ1or by addition of
FKBP52 directly to ðHsp90Þ2, yields similar KDvalues of 140 nM
and 145 nM respectively. This data not only shows that the subse-
quent binding of a TPR-containing cochaperone is independent of
The KDvalues obtained for Hsp70 binding to the three Hsp90∕
Hop complexes yields low nanomolar values affirming our method
and implying a favorable binding event for Hsp70 (KD¼ 20 nM)
following the binding of Hop to Hsp90.
The functionality of the Hsp90 machinery is known to depend on
a number of chaperones and cochaperones to drive the cycle. The
precise order of binding and the stoichiometries of the various
complexes formed however have remained controversial. Using
MS we have explored all possible complexes that can be formed
by four proteins known to have critical roles in the Hsp90 chaper-
one cycle: Hsp90, Hsp70, Hop, and FKBP52. Initially, we found
that Hsp90 was predominantly dimeric and FKBP52 monomeric,
both were anticipated. Recently there has been some controversy
over the oligomeric state of Hop reported as dimeric (6, 20, 34)
under some conditions (36). We have shown that under the con-
ditions used here both proteins are monomeric.
To characterize the complexes formed between the compo-
nents of the Hsp90 machinery we incubated different ratios of
the four proteins, varying both their concentration and order of
addition. The results enabled us to construct a network of pos-
sible complexes (Fig. 4). Despite the fact that all proteins are at
micromolar concentration in solution, Hsp90 exists in monomeric
and dimeric forms. Each Hsp90 monomer possesses a C-terminal
MEEVD motif for TPR protein binding. We found that some
complexes, which one might expect to form in theory, such as the
ðHsp90Þ1ðHopÞ1complex in practice did not form. For the 10 dif-
ferent species of Hsp90 and cochaperones complexes observed,
we determined their binding stoichiometry as well as the likely
assembly pathway. Starting with Hsp90dimers (green oval Fig. 4),
up to two Hop monomers can bind to form ðHsp90Þ2ðHopÞ1and
ðHsp90Þ2ðHopÞ2complexes (Fig. 1 A and B). The absence of the
ðHsp90Þ1ðHopÞ1complex was in contrast to the results obtained
for FKBP52, wherein the complex ðHsp90Þ1ðFKBP52Þ1was
observed. Our observation implies that dimerization of Hsp90 is
critical for Hop but not for FKBP52 binding, consistent with the
15´Å resolution EM structure of the Hsp90-Hop complex in which
each Hop monomer interacts with both subunits of ðHsp90Þ2(37).
The binding affinities of FKBP52 and Hop for the two binding
sites in the Hsp90 dimer, as well as additional binding of Hsp70,
are all in the nanomolar range. However, there is a wide varia-
bility in the KDvalues reported for binary interactions between
these proteins, both for human and yeast proteins (5, 6, 38)
(SI Appendix, Table S3 and Fig. S10). This variability could arise
from possible differences between the protein homologues, the
varied solution conditions, and methods employed in their mea-
surement. An apparent advantage of our method is that it gives a
consistent set of KDvalues for a system of complexes, using the
same method and solution conditions. Our data allows a direct
comparison of the values determined for the different complexes
formed in the cycle. Overall, our KDvalues are somewhat lower
than those measured in an earlier study for C-terminal peptides
of Hsp90 binding to isolated TPR domains (39) but are however
very similar to some published values, where available, for the
full-length human proteins (7, 40) (see SI Appendix, Table S3
and Fig. S10). Our KDvalues of 15 nM and 55 nM for the binding
of the first and second Hop molecule are in accord with the
(A) MS of an intermediate complex ðHsp90Þ2ðHopÞ1ðHsp70Þ1(pink) formed
after addition of Hsp70 following Hop binding to the ðHsp90Þ2. B) Addition
of FKBP52 to the intermediate complex to form ðHsp90Þ2ðHopÞ1ðFKBP52Þ1
ðHsp70Þ1(olive green), (SI Appendix, Figs. S6 and S7 for additional quaternary
Hsp70 binds to ðHsp90Þ2and Hop to form an intermediate complex.
www.pnas.org/cgi/doi/10.1073/pnas.1106261108 Ebong et al.
proposal that binding of the first Hop results in a subtle decrease
in affinity of the Hsp90 dimer for the second Hop (37).
The KDvalues for the binding of FKBP52 and Hop to either
site in the ðHsp90Þ2are similar. We speculate that the slight pre-
ference for Hop over FKBP52 binding may be due to the addi-
tional interactions Hop makes with the C-terminal and middle
domain of Hsp90, as previously established using ITC (20) and
more recently revealed by the Hsp90-Hop EM structure (37).
Despite its slightly less favorable KD, FKBP52 is however able to
form an asymmetric complex with Hop and Hsp90 (Fig. 1B). This
finding is in accord with the recent observation that the ternary
complex is preferentially formed (22). The asymmetric complex
observed here forms regardless of the order of addition of either
FKBP52 or Hop and is statistically favored and stable for at least
3 h under these conditions.
It has been proposed that Hop binds directly to Hsp70 with
ADP present (7, 41). However, other studies have shown that
Hop can bind to Hsp70 in the absence of nucleotide (42). Our
results show evidence of a weak binding between Hsp70 and
Hop but only when ADP is present (KD¼ 6 μM). Under the con-
ditions used here, no binding was observed for Hsp70 and Hop
in the absence of nucleotide. Favorable binding was however
observed after prior binding of Hop to the Hsp90 dimer (KD¼
20 nM). Our observation suggests a mechanism in which Hop
and/or Hsp90 undergo a conformational change upon binding
which renders the complex competent to bind Hsp70, and agrees
with the recent EM structure where Hop binds a distinct con-
formation of Hsp90 which lies somewhere between the open con-
formation, observed for the apo state, and the N-terminally
dimerized closed state (37). We find that the number of Hsp70
molecules that bind to the various Hsp90 assemblies is directly
related to the number of Hop molecules within the complex.
We also observe a complex with all four components ðFKBP52Þ1
ðHsp90Þ2ðHopÞ1ðHsp70Þ1proposed previously (22) and discov-
ered experimentally here. Interestingly, our KDvalues show that
binding of Hsp70 is not affected by prior binding of FKBP52,
Hop, or Hop and Hsp70 to the other binding site. Considering
the cellular concentrations recently reported as 10 μM for Hsp70,
5 μM ðHsp90Þ2, and 3 μM for Hop (43), even though two Hop
monomers can bind to an Hsp90 dimer the ðHsp90Þ2ðHopÞ1com-
plex is likely the physiologically relevant species. It has also been
shown recently that a single Hop molecule is sufficient to com-
pletely inhibit the ATPase activity of Hsp90 (22). The ðHsp90Þ2
ðHopÞ1complex in turn can bind to Hsp70 to form the inter-
mediate complex, that binds FKBP52 to form the functional
chaperone complex (Fig. 4).
Given the ratios of ðHsp90Þ2to Hop and FKBP52 (10∶1 and
15∶1 respectively) in the cell (44), it is perhaps surprising that the
asymmetric complex is able to form by addition of an FKBP52
molecule. However, estimated concentrations do not take into
account elevated protein levels in response to the stress environ-
ment, or the effects of crowding within the cell (45). The asym-
metric complex is, however, statistically more likely to form than
the ðHsp90Þ2ðHopÞ2complex if concentrations of FKBP52 and
Hop are comparable. Our results do show however that no addi-
tional FKBP52 molecule can bind to the ðHsp90Þ2ðHopÞ2com-
plex (Fig. 1), or those containing two Hop molecules with one or
two Hsp70 molecules (SI Appendix, Fig. S7). Because FKBP52
is always present as a final component of the mature receptor
complex (14), our data implies that binding of two Hop molecules
renders the complex incompetent of further development into a
functional chaperone complex. Our data is in contrast to earlier
studies in which ðHsp90Þ2binding to dimeric Hop was considered
to be a prerequisite for chaperone function (4, 6, 7) but is con-
sistent with the proposal that the asymmetric nature of Hsp90-
Hop complexes are important in the Hsp90 cycle (22, 37).
The simultaneous binding of Hsp70 and Hsp90 to Hop has
been proposed to bring the two proteins together into an active
complex in which client proteins are transferred from Hsp70
to Hsp90 to complete their folding and maturation (15). Given
the cellular concentration of Hsp70, which is higher than both
ðHsp90Þ2and Hop, the favorable KDfor Hsp70 (20 nM) binding
would ensure that the ðHsp90Þ2ðHopÞ1complexes once formed
can proceed to productive and competent folding complexes by
binding FKBP52 to form ðHsp90Þ2ðHopÞ1ðFKBP52Þ1ðHsp70Þ1.
By contrast the complexes of ðHsp90Þ2ðHopÞ2ðHsp70Þ1and
ðHsp90Þ2ðHopÞ2ðHsp70Þ2, if formed, are incapable of subsequent
FKBP52 binding and are therefore unlikely to play any further
role in the chaperone cycle.
In summary, by incubating Hsp90 with different cochaperones,
and exploring all possible reaction products, we have determined
the dominant complexes formed during the Hsp90 cycle. We have
FKBP52 KDS for Hop, ðHsp90Þ2
multiple spectra from eight differ-
ent experimental datasets, includ-
ing replicate experiments (green
background). Favorable KDs are
represented with thicker arrows.
Addition of Hsp70 to the ternary
plexes leads to the formation of
quaternary complexes with favor-
able KDvalues (blue background).
Complexes likely to form, given
the cellular concentrations of the
circled (dotted line). KDvalues and
associated errors are shown with
values deduced form the same
dataset represented by the same
An interaction network
Hsp90,Hsp70, Hop, and
Ebong et al.PNAS
November 1, 2011
also identified some key binding events that likely lead to the
assembly of productive folding machinery in the Hsp90 chaper-
one cycle. The similarity of the KDvalues for the ten species
studied here ensure that in the cellular environment the chaper-
one pathway will be populated with complexes that are both
heterogeneous and dynamic and which can likely be finely tuned
to respond to the stress conditions experienced. More generally,
our results highlight the distinct advantages of MS in being able
to determine simultaneously, and in real time, the composition
and relative abundance of each complex in a highly heteroge-
neous system from many different datasets. This approach should
have broad application to other systems as it not only gives in-
sights into the interactions, but also leads to a self-consistent set
of KDvalues as shown here for the different complexes formed in
the Hsp90 chaperone cycle.
Protein Purification. Hsp90 and cochaperones were human, expressed in
Escherichia coli and purified as previously described (20, 29). Prior to MS
analysis, 20 μL of the protein solutions were buffer exchanged into 50 mM
ammonium acetate (AmAc), pH 7.5 at 4°C using micro Bio-Spin® columns
Binding Studies. Proteins were combined to give a range of concentrations
from 1 to 5 μM in 50 mM AmAc pH 7.5 and incubated for 30 min on ice.
For Hsp90 and Hop binary complexes were formed in 50 mM AmAc pH 7.5.
As a control all complexes were formed in a standard buffer 50 mM Tris HCl
pH 7.5. MS showed no difference in the ratios of the complexes formed.
Competition experiments were performed by preforming binary complexes
of Hsp90 with Hop or FKBP52 in 50 mM AmAc and adding FKBP52 or Hop
immediately prior to MS. For nucleotide studies, ADP, AMPpNp, and ATP
were prepared at 1 mM in 50 mM AmAc, pH 7.5 and added to give a final
solution concentration of 100 μM and 300 μM nucleotide with an equivalent
MS. Spectra were obtained on a QToF II MS (Waters) modified for studying
noncovalent interactions (27, 46). 2.5 μL aliquots were introduced into the MS
from a gold-plated capillary needle (Harvard Apparatus). Capillary voltages
were 1.5–1.8 kV, cone voltages from 80–130 V, and a collision voltages were
from 50–150 V. Spectra were calibrated externally using caesium iodide.
ACKNOWLEDGMENTS. We thank funding from the European Union Prospects
grant number (HEALTH-F4-2008-201648) (to N.M.) and Engineering and
Physical Sciences Research Council for funding (to I.E.) Faculty of science
and technology (FCT), Portugal for funding (to M.A.S.) and British Petroleum
Centenary Murray Edwards College Fund and Cambridge Commonwealth
Trust for funding (to S.D.). C.V.R. is a Royal Society Professor.
1. Pratt WB, Dittmar KD (1998) Studies with purified chaperones advance the under-
standing of the mechanism of glucocorticoid receptor-hsp90 heterocomplex assembly.
Trends Endocrin Met 9:244–252.
2. Richter K, Buchner J (2001) Hsp90: chaperoning signal transduction. J Cell Physiol
3. Picard D (2002) Heat-shock protein 90, a chaperone for folding and regulation. Cell
Mol Life Sci 59:1640–1648.
4. Siligardi G, et al. (2002) Regulation of Hsp90 ATPase activity by the co-chaperone
Cdc37p/p50(cdc97). J Biol Chem 277:20151–20159.
5. Siligardi G, et al. (2004) Co-chaperone regulation of conformational switching in the
Hsp90 ATPase cycle. J Biol Chem 279:51989–51998.
6. Prodromou C, et al. (1999) Regulation of Hsp90 ATPase activity by tetratricopeptide
repeat (TPR)-domain co-chaperones. EMBO J 18:754–762.
7. Hernandez MP, Sullivan WP, Toft DO (2002) The assembly and intermolecular
properties of the hsp70-Hop-hsp90 molecular chaperone complex. J Biol Chem 277:
8. Pearl LH, Prodromou C (2006) Structure and mechanism of the Hsp90 molecular
chaperone machinery. Annu Rev Biochem 75:271–294.
9. Obermann WMJ, Sondermann H, Russo AA, Pavletich NP, Hartl FU (1998) In vivo func-
tion of Hsp90 is dependent on ATP binding and ATP hydrolysis. J Cell Biol 143:901–910.
10. Grenert JP, Johnson BD, Toft DO (1999) The importance of ATP binding and hydrolysis
by hsp90 in formation and function of protein heterocomplexes. J Biol Chem 274:
11. Krukenberg KA, Street TO, Lavery LA, Agard DA (2011) Conformational dynamics of
the molecular chaperone Hsp90. Q Rev Biophys 1:1–27.
12. Southworth DR, Agard DA (2008) Species-dependent ensembles of conserved confor-
mational states define the Hsp90 chaperone ATPase cycle. Mol Cell 32:631–640.
13. Owens-Grillo JK, et al. (1996) A model of protein targeting mediated by immunophi-
lins and other proteins that bind to hsp90 via tetratricopeptide repeat domains. J Biol
14. Pratt WB, Toft DO (1997) Steroid receptor interactions with heat shock protein and
immunophilin chaperones. Endocr Rev 18:306–360.
15. Chen SY, Smith DF (1998) Hop as an adaptor in the heat shock protein 70 (Hsp70) and
Hsp90 chaperone machinery. J Biol Chem 273:35194–35200.
16. D’Andrea LD, Regan L (2003) TPR proteins: the versatile helix. Trends Biochem Sci
17. Scheufler C, et al. (2000) Structure of TPR domain-peptide complexes: critical elements
in the assembly of the Hsp70-Hsp90 multichaperone machine. Cell 101:199–210.
18. Morishima Y, et al. (2000) The Hsp organizer protein Hop enhances the rate of but
is not essential for glucocorticoid receptor folding by the multiprotein Hsp90-based
chaperone system. J Biol Chem 275:6894–6900.
19. Richter K, Muschler P, Hainzl O, Reinstein J, Buchner J (2003) Sti1 is a non-competitive
inhibitor of the Hsp90 ATPase—binding prevents the N-terminal dimerization reaction
during the ATPase cycle. J Biol Chem 278:10328–10333.
20. Onuoha SC, Coulstock ET, Grossmann JG, Jackson SE (2008) Structural studies on the
co-chaperone Hop and its complexes with Hsp90. J Mol Biol 379:732–744.
21. HildenbrandZL, et al. (2011) Hsp90 can accommodatethe simultaneous binding of the
FKBP52 and Hop proteins. Oncotarget 2:45–58.
22. Li J, Richter K, Buchner J (2011) Mixed Hsp90—cochaperone complexes are important
for the progression of the reaction cycle. Nat Struct Mol Biol 18:61–66.
23. Young JC, Obermann WMJ, Hartl FU (1998) Specific binding of tetratricopeptide
repeat proteins to the C-terminal 12-kDa domain of hsp90. J Biol Chem 273:
24. Radanyi C, Chambraud B, Baulieu EE (1994) The ability of the immunophilin FKBP59-
HBI to interact with the 90-kDa heat shock protein is encoded by its tetratricopeptide
repeat domain. Proc Natl Acad Sci USA 91:11197–11201.
25. Sharon M, Robinson CV (2007) The role of mass spectrometry in structure elucidation
of dynamic protein complexes. Annu Rev Biochem 76:167–193.
26. Stengel F, et al. (2010) Quaternary dynamics and plasticity underlie small heat shock
protein chaperone function. Proc Natl Acad Sci USA 107:2007–2012.
27. Hernandez H, Robinson CV (2007) Determining the stoichiometry and interactions of
macromolecular assemblies from mass spectrometry. Nat Protoc 2:715–726.
28. Karagoz GE, et al. (2011) N-terminal domain of human Hsp90 triggers binding to the
cochaperone p23. Proc Natl Acad Sci USA 108:580–585.
29. McLaughlin SH, Smith HW, Jackson SE (2002) Stimulation of the weak ATPase activity
of human Hsp90 by a client protein. J Mol Biol 315:787–798.
30. McLaughlin SH, Ventouras LA, Lobbezoo B, Jackson SE (2004) Independent ATPase ac-
tivity of Hsp90 subunits creates a flexible assembly platform. J Mol Biol 344:813–826.
31. Prodromou C, et al. (1997) Identification and structural characterization of the ATP/
ADP-binding site in the Hsp90 molecular chaperone. Cell 90:65–75.
32. Hernandez H, et al. (2009) Isoforms of U1-70k control subunit dynamics in the human
spliceosomal U1 snRNP. PLoS One 4:e7202.
33. Lane LA, et al. (2011) Mass spectrometry reveals stable modules in holo and apo RNA
polymerases I and III. Structure 19:90–100.
34. Flom G, Behal RH, Rosen L, Cole DG, Johnson JL (2007) Definition of the minimal frag-
ments of Sti1 required for dimerization, interaction with Hsp70 and Hsp90 and in vivo
functions. Biochem J 404:159–167.
35. Yi F, Doudevski I, Regan L (2010) HOP is a monomer: investigation of the oligomeric
state of the co-chaperone Hop. Protein Sci 19:19–25.
36. Benaroudj N, Batelier G, Triniolles F, Ladjimi MM (1995) Self-association of the mole-
cular chaperone HSC70. Biochemistry 34:15282–15290.
37. Southworth DR, Agard DA (2011) Client-loading conformation of the Hsp90 molecular
chaperone revealed in the Cryo-EM structure of the human Hsp90∶Hop complex.
Mol Cell 42:771–781.
38. Wandinger SK, Richter K, Buchner J (2008) The Hsp90 chaperone machinery. J Biol
39. Cortajarena AL, Regan L (2006) Ligand binding by TPR domains. Protein Sci 15:
40. Pirkl F, Buchner J (2001) Functional analysis of the Hsp90-associated human peptidyl
prolyl Cis/Trans isomerases FKBP51, FKBP52 and cyp40. J Mol Biol 308:795–806.
41. Johnson BD, Schumacher RJ, Ross ED, Toft DO (1998) Hop modulates hsp70∕hsp90
interactions in protein folding. J Biol Chem 273:3679–3686.
42. Wegele H, Haslbeck M, Reinstein J, Buchner J (2003) Sti1 is a novel activator of the Ssa
proteins. J Biol Chem 278:25970–25976.
43. Kundrat L, Regan L (2010) Balance between folding and degradation for Hsp90-
dependent client proteins: a key role for CHIP. Biochemistry 49:7428–7438.
44. Ghaemmaghami S, et al. (2003) Global analysis of protein expression in yeast. Nature
45. Ellis RJ, Minton AP (2006) Protein aggregation in crowded environments. Biol Chem
46. Sobott F, Hernández H, McCammon MG, Tito MA, Robinson CV (2002) A tandem
mass spectrometer for improved transmission and analysis of large macromolecular
assemblies. Anal Chem 74:1402–1407.
www.pnas.org/cgi/doi/10.1073/pnas.1106261108 Ebong et al.