Available via license: CC BY 4.0
Content may be subject to copyright.
Roles of Individual Domains and Conserved Motifs of the AAAⴙ
Chaperone ClpB in Oligomerization, ATP Hydrolysis, and
Chaperone Activity*
Received for publication, September 20, 2002, and in revised form, January 23, 2003
Published, JBC Papers in Press, March 6, 2003, DOI 10.1074/jbc.M209686200
Axel Mogk‡§¶, Christian Schlieker‡, Christine Strub§, Wolfgang Rist‡, Jimena Weibezahn‡,
and Bernd Bukau‡储
From the ‡Zentrum fu¨ r Molekulare Biologie Heidelberg, Im Neuenheimer Feld 282, Heidelberg D-69120, Germany and
§Institut fu¨ r Biochemie und Molekularbiologie, Universita¨ t Freiburg, Hermann-Herder-Strasse 7,
79104 Freiburg, Germany
ClpB of Escherichia coli is an ATP-dependent ring-
forming chaperone that mediates the resolubilization of
aggregated proteins in cooperation with the DnaK chap-
erone system. ClpB belongs to the Hsp100/Clp subfamily
of AAAⴙproteins and is composed of an N-terminal do-
main and two AAA-domains that are separated by a
“linker” region. Here we present a detailed structure-
function analysis of ClpB, dissecting the individual roles
of ClpB domains and conserved motifs in oligomeriza-
tion, ATP hydrolysis, and chaperone activity. Our re-
sults show that ClpB oligomerization is strictly depend-
ent on the presence of the C-terminal domain of the
second AAA-domain, while ATP binding to the first AAA-
domains stabilized the ClpB oligomer. Analysis of mu-
tants of conserved residues in Walker A and B and sen-
sor 2 motifs revealed that both AAA-domains contribute
to the basal ATPase activity of ClpB and communicate in
a complex manner. Chaperone activity strictly depends
on ClpB oligomerization and the presence of a residual
ATPase activity. The N-domain is dispensable for oli-
gomerization and for the disaggregating activity in vitro
and in vivo. In contrast the presence of the linker region,
although not involved in oligomerization, is essential for
ClpB chaperone activity.
The Escherichia coli chaperone ClpB belongs to the ring-
forming Clp/Hsp100 proteins. Clp/Hsp100 proteins can be clas-
sified into two distinct subfamilies. Class I proteins (ClpA and
ClpB in E. coli) are composed of two highly conserved nucleo-
tide binding domains (termed ATP-1 and ATP-2), whereas class
II proteins (ClpX and HslU, as representatives of E. coli) con-
tain only a single NBD (homologous to ATP-2) (1). Sequence
analysis of the NBDs revealed a significant sequence homology
between Clp/Hsp100 and AAA proteins (ATPase associated
with a variety of cellular activities), and consequently a new
AAA⫹superfamily, representing both protein classes, was pro-
posed (2). The structural basis of this superfamily was con-
firmed by determination of the first Clp/Hsp100 protein struc-
ture, HslU, that showed significant similarity to the AAA
proteins N-ethylmaleimide-sensitive fusion protein and p97 (3,
4). The recently solved crystal structure of the first nucleotide
binding domain of ClpB also demonstrated the close structural
relationship between Clp/Hsp100 and AAA proteins (5). The
conserved AAA-domain (also referred to as AAA module) is
made up of two domains, a core region that forms the nucleo-
tide binding pocket, containing the classical Walker A and B
motifs, and a C-terminal
␣
-helical domain (C-domain). The
ATP binding pocket is located at the interface of neighboring
subunits in the oligomer. The C-domain contacts its own core
ATPase domain and that of adjacent subunits and is involved
in nucleotide binding and hexamerization in HslU (3, 4). Be-
sides sensing the nucleotide status of the core ATPase domain,
C-terminal domains of the second AAA-domain have also been
proposed to mediate substrate interaction and were therefore
termed the sensor and substrate discrimination (SSD)
1
do-
mains (6).
In addition Hsp100/Clp proteins contain variable regions at
their N terminus. ClpA and ClpB have homologous N-domains
of about 150 residues that consist of two sequence repeats and
form an independent structural domain still of unknown func-
tion (7). ClpX possesses a zinc binding domain at the N termi-
nus (8), whereas HslU lacks an N-terminal domain but rather
contains an extra domain (the I-domain) inserted into the AAA-
domain (3, 4). The most striking difference between members
within the class I subfamily is the presence/absence of a region
that is proposed to link the two AAA-domains ATP-1 and
ATP-2. The presence of this variable linker region serves as a
criteria for classification of Hsp100 proteins; the linker is long-
est in ClpB (⬃140 residues) but is absent in ClpA (see Fig. 1)
(1).
ClpB is unique among the Hsp100/Clp proteins because it
does not associate with a proteolytic partner protein. Recently,
an essential docking site of the peptidase ClpP was identified
within ClpX (9). The signature motif (LIV-G-FL) is conserved
in all other ClpP-interacting proteins like ClpA but is missing
in ClpB, thereby explaining why ClpB acts independently of
peptidases. Instead ClpB mediates the resolubilization of ag-
gregated proteins in cooperation with the DnaK chaperone
system (10 –13). The mechanism of the disaggregation reaction
and the basis of ClpB/DnaK cooperation are still not
understood.
Here we report a structure-function analysis of ClpB that is
* This work was supported by Deutsche Forschungsgemeinschaft
Grant Bu617/14-1 and by the Fond der Chemischen Industrie (to B. B.).
The costs of publication of this article were defrayed in part by the
payment of page charges. This article must therefore be hereby marked
“advertisement” in accordance with 18 U.S.C. Section 1734 solely to
indicate this fact.
¶To whom correspondence may be addressed. E-mail: a.mogk@
zmbh-uni-heidelberg.de.
储To whom correspondence may be addressed. E-mail: bukau@
zmbh.uni-heidelberg.de.
1
The abbreviations used are: SSD, sensor- and substrate discrimina-
tion; DTT, dithiothreitol; ATP
␥
S, adenosine 5⬘-O-(thiotriphosphate);
IPTG, isopropyl-1-thio-

-D-galactopyranoside; MDH, malate dehydro-
genase; WT, wild type; aa, amino acid.
THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 278, No. 20, Issue of May 16, pp. 17615–17624, 2003
© 2003 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.
This paper is available on line at http://www.jbc.org 17615
by guest on December 27, 2015http://www.jbc.org/Downloaded from
aimed at identifying the roles of individual domains and con-
served motifs in the disaggregation process and the coupling of
the ATPase cycle with the chaperone activity. Constructed
ClpB variants were characterized with respect to their struc-
tural integrity, as determined by oligomerization studies and
partial proteolysis. Additionally the ATPase and chaperone
activities of all constructs were tested.
EXPERIMENTAL PROCEDURES
Strains and Plasmids—E. coli strains used were derivatives of
MC4100 (araD139 ⌬(argF-lac)U169 rpsL150 relA1 flbB5301 deoC1
ptsF25 rbsR). clpB mutant allele was introduced by P1 transduction
into MC4100 background to generate strains BB4561 (⌬clpB::kan). The
E. coli clpB gene was cloned by PCR into the pUHE21 expression vector
and verified by DNA sequencing. Mutants and deletion variants were
generated by using standard PCR techniques. ClpB-(⌬410 –532) was
constructed by replacing the entire linker region with an SpeI site,
leading to the insertion of two amino acids (Thr and Ser) at the deletion
site. Reinsertion of the linker was obtained by PCR amplification of the
corresponding region (amino acid 410 –530) and addition of an SpeI site
5⬘and an NheI site 3⬘to the linker fragment. This fragment was
digested with NheI/SpeI and inserted into the SpeI site of the ⌬410 –532
construct, leading to the insertion of two amino acids (Thr and Ser or
Ala and Ser) at each domain boundary (ClpB ⌬LBL). Each mutagenesis
was confirmed by DNA sequencing.
Proteins—Wild type and mutant ClpB were purified as described
after overproduction in ⌬clpB::kan cells (11). Purifications of DnaK,
DnaJ, and GrpE were according to published protocols (14). Pyruvate
kinase and
␣
-casein were purchased from Sigma; malate dehydrogen-
ase (MDH) was from pig heart muscle, and firefly luciferase from Roche
Applied Science. Protein concentrations were determined with the Bio-
Rad Bradford assay using bovine serum albumin as standard. Protein
concentrations refer to the protomer.
Tryptophan Fluorescence—Measurements of the intrinsic trypto-
phan fluorescence of ClpB were performed on a PerkinElmer Life Sci-
ences LS50B spectrofluorimeter. The emission spectra of tryptophan
fluorescence of ClpB (0.5
M) in the absence of nucleotide or the pres-
ence of 2 mMATP/ADP were recorded at 30 °C in buffer A (50 mMTris,
pH 7.5, 150 mMKCl, 20 mMMgCl
2
,2mMDTT) between 300 and 400 nm
at a fixed excitation wavelength of 290 nm.
ATPase Activity Assay—ATP hydrolysis rates under steady-state
conditions were determined as described (14). Reactions were per-
formed at 30 °C in buffer A (50 mMTris, pH 7.5, 150 mMKCl, 20 mM
MgCl
2
,2mMDTT) containing 0.5
MClpB (wild type or derivatives), 2
mMATP, and [
␣
-
32
P]ATP (0.1
Ci, Amersham Biosciences). ATPase
activities were also determined in the presence of 0.25 mg/ml
␣
-casein
or 100 mMammonium sulfate. Hydrolysis was quantified by using the
program MACBAS version 2.5 (Fuji), and rates of ATP hydrolysis were
determined by using the program GRAFIT version 3.0 (Erithacus
software).
Size Exclusion Chromatography—Size exclusion chromatography
was performed at room temperature in buffer A containing 5% (v/v)
glycerol. Nucleotide-dependent oligomerization was followed in the
presence of 2 mMATP or ADP in the running buffer. 10
MClpB was
incubated in buffer A in the absence or presence of nucleotides (2 mM
ATP/ADP) for 5 min at room temperature, followed by injection into the
high pressure liquid chromatography system (PerkinElmer Life Sci-
ences) connected to a SEC 400 column (Bio-Rad). Chromatographic
steps were performed with a flow rate of 0.8 ml/min.
Cross-linking Assays—All ClpB variants were dialyzed against
buffer B (50 mMHEPES, pH 7.5, 150 mMKCl, 20 mMMgCl
2
,2mM
DTT). 1
MClpB was incubated at 30 °C in the absence or presence of
nucleotides (2 mMATP or ADP) for 5 min. Cross-linking reactions were
started by addition of 0.1% glutaraldehyde and incubated for another
10 min. Reactions were stopped by addition of 1 MTris, pH 7.5, and
cross-linking products were analyzed by SDS-PAGE (4 –15%) followed
by silver staining.
Partial Proteolysis and Identification of Cleavage Products—1
M
ClpB (wild type or derivative) was incubated at 30 °C in buffer A
without DTT for 5 min in the absence or presence of nucleotide (2 mM
ATP, ADP, and ATP
␥
S). Proteolysis was initiated upon addition of 0.2
g/ml thermolysin or subtilisin, and generated cleavage products were
analyzed by SDS-PAGE (15%) and silver staining. Kinetic analysis of
the degradation reaction revealed the occurrence of stable fragments
after 30 –60 min of incubation time. Identity of cleavage products was
determined by N-terminal sequencing (TopLab) and mass spectrome-
try. For mass spectrometry analysis, bands were excised from one-
dimensional Coomassie Silver-stained SDS-polyacrylamide gels and
in-gel-digested with trypsin as described (15). Tryptic peptides were
analyzed by nanoelectrospray tandem mass spectrometry as described
previously (16) using a QSTAR
TM
Pulsar (MDS Sciex, Toronto, Canada)
equipped with a nanoelectrospray ion source (MDS Proteomics, Odense,
Denmark). Sequence searches were performed with the Protein and
Peptide Software Suite (MDS Proteomics).
In Vitro Activity Assays—2
MMDH was denatured in buffer A for
30 min at 47 °C. Turbidity of 1
Maggregated MDH was measured
at 30 °C at an excitation and emission wavelength of 550 nm
(PerkinElmer Life Sciences luminescence spectrometer LS50B). De-
crease of light scattering was followed upon addition of 1.5
MClpB
(wild type or derivative), the DnaK chaperone system (1
MDnaK, 0.2
MDnaJ, 0.1
MGrpE), and an ATP-regenerating system (3 mM
phosphoenolpyruvate, 20 ng/ml pyruvate kinase, 2 mMATP). Disag-
gregation rates were derived from the linear decrease of turbidity
between 5 and 30 min.
0.2
Mluciferase was denatured in buffer A for 15 min at 43 °C.
Refolding of 0.1
Maggregated luciferase was initiated by addition of 1
MClpB (wild type or derivative), the DnaK system (0.5
MDnaK, 0.1
MDnaJ, 0.05
MGrpE), and an ATP-regenerating system. Luciferase
activities were determined as described, and refolding rates were cal-
culated from the linear increase of luciferase activities between 15 and
90 min.
In Vivo Activity Assays—Survival of E. coli cells after exposure to
lethal temperature was determined by calculating the plating effi-
ciency. Cells were grown in LB medium to mid-exponential growth
phase at 30 °C followed by incubation at 50 °C for the indicated time.
Serial dilutions (10
⫺3
to 10
⫺7
) of cells were prepared in LB medium and
spotted onto LB plates. Plates were incubated for 24 h at 30 °C, and
colony numbers were determined afterward. Defects in protein disag-
gregation in ⌬clpB::kan mutant cells were followed by isolation of
aggregated proteins as described (17).
RESULTS
Design of ClpB Mutants and Deletions Variants—ClpB is
proposed to consist of an N-terminal domain, two AAA-domains
(ATP-1 and ATP-2), which are separated by a linker region, and
the C-terminal SSD domain (as illustrated in Fig. 1). In order
to probe the proposed domain organization of ClpB, several
deletion variants were constructed, based on sequence and
secondary structure analysis of ClpB and its comparison with
AAA⫹proteins of known structure. Due to the presence of an
internal start codon, two versions of E. coli ClpB exist in vivo:
a full-length protein (aa 1–857) and an N-terminally trun-
cated variant (aa 149 –857). Both versions have been demon-
strated to form mixed oligomers (18). In order to work with
uniform proteins, we decided to mutate the internal start
codon of wild type ClpB, leading to the production of full-
length ClpB only. This ClpB derivative exhibited in vitro and
in vivo chaperone activity, indistinguishable from ClpB ex-
pressed from the wild type gene (data not shown). The cor-
responding construct was used as basis for the construction of
all ClpB derivatives.
The border between the N-domain and the first AAA-domain
is often characterized by the presence of an internal start codon
within the clpB gene (Val-149). Sequence analysis of Hsp100
N-domains and secondary structure prediction of 15 different
ClpB proteins revealed that the ClpB N-domain is built up of
residues 1–144 (7) (data not shown). We therefore decided to
use Met-143 as the start site for ClpB ⌬N-(143–857) version.
The C-terminal SSD was removed in the ClpB ⌬SSD variant
(aa 1–758). AAA-domains were suggested to be separated by
the insertion of the linker region within ClpB. Recent sequence
and structural analysis of AAA⫹proteins revealed that the
linker region potentially interrupts the first AAA-domain in-
stead of separating both AAA modules (2). Several ClpB dele-
tion variants were constructed to test this possibility: ClpB-
(1–409), ClpB-(551–857), and ClpB-(1–567).
Several conserved motifs and residues are proposed to be
involved in ATP binding and hydrolysis by AAA proteins. The
highly conserved Lys residues of the Walker A motif (Lys-212
Structure-Function Analysis of ClpB17616
by guest on December 27, 2015http://www.jbc.org/Downloaded from
and Lys-611 of ClpB) contact the phosphate groups of the

and
␥
nucleotides in the first and second AAA-domain, respectively
(19). The conserved Glu residues of the Walker B motif (Glu-
279 and Glu-678 of ClpB) are proposed to represent the cata-
lytic base for ATP hydrolysis in the first and second AAA-
domain, respectively. AAA⫹proteins additionally contain a
conserved Arg residue, termed sensor 2 (2). Sensor 2 lies in the
C-terminal domain of each AAA module and contacts the
␥
-phosphate of bound ATP. This Arg residue is part of an
invariant GAR motif (
813
GAR
815
in ClpB) within the SSD do-
main of AAA⫹(Hsp100) proteins. It is proposed that the con-
served arginine can sense the nucleotide status and mediate
conformational changes of the C-terminal domain relative to
the core domain during ATP hydrolysis (19, 20).
The formation of hexameric rings in AAA⫹proteins brings
residues of adjacent protomers into close proximity to ATP,
bound to a neighboring subunit. Such interactions could pro-
vide the basis for an intermolecular catalytic mechanism, re-
sulting in cooperative ATP hydrolysis within the oligomer.
Previous studies on the AAA protein FtsH have demonstrated
that Arg residues (Arg-332 and Arg-756 in ClpB) can poten-
tially act as “Arg fingers”by contacting the ATP bound to a
neighbored subunit (21). All described motifs, involved in ATP
binding and hydrolysis, were subjected to alanine mutagenesis,
as illustrated in Fig. 1. ClpB mutants and deletion variants
were purified and characterized with respect to oligomeriza-
tion, ATP hydrolysis, and chaperone activity.
Oligomerization of ClpB Variants—ClpB proteins form hex-
americ ring structures in the presence of ATP (22). Oligomer-
ization of ClpB mutants and deletion variants was followed by
size exclusion chromatography in the absence or presence of
ATP and ADP. Consistent with published data, ClpB is only
capable of hexamerization in the presence of ATP (Fig. 2A). In
the absence of nucleotide, ClpB eluted predominantly as a
dimer, although ADP addition led to the formation of trimers
and/or tetramers. Analysis of ClpB deletion variants revealed
that isolated AAA-domains (1–409, 1–567, and 551–857) can-
not oligomerize. In agreement with previous data (23), hexa-
merization is strictly dependent on the presence of the SSD
domain, although the N-domain and the linker region are dis-
pensable (Table I). Characterization of ClpB point mutants
revealed that besides the C-terminal SSD domain, the first
ATPase domain also contributes to oligomerization: mutations
in the Walker A motif (K212A) or the Arg finger (R332A) of the
first AAA-domain abolished hexamerization, although corre-
sponding mutations in the second AAA module (K611A and
R756A) domain had no effect (Fig. 2Band Table I). Finally, all
Walker B mutants in the two AAA modules (E279A, E678A,
and E279A/E678A), as well as the sensor 2 mutant (
813
AAA
815
),
did not exhibit oligomerization defects.
In an additional approach, oligomerization characteristics
were studied by glutaraldehyde cross-linking. Fast cross-link-
ing of ClpB monomers to oligomeric species was obtained
within 10 min in the absence of nucleotide, indicating that
ClpB assembly can in principle also occur without nucleotide.
Because such nucleotide-independent hexamerization was not
observed during gel filtration runs, the formed oligomers seem
to be unstable in the absence of ATP.
Kinetic analysis of the cross-linking reaction revealed that a
ladder of cross-linking products preceded the formation of the
fully cross-linked oligomer (data not shown). The presence of
ATP or ADP accelerated the cross-linking reaction and also
slightly changed the size of the fully cross-linked oligomer.
Whereas the ladder of cross-link products could be followed up
to a heptamer in the absence of nucleotide, addition of ATP and
ADP resulted predominantly in cross-linking to the hexameric
and pentameric species, respectively (Fig. 3A). The existence of
heptameric ClpB rings have also been shown by Chung and
co-workers (18) using electron microscopy.
Cross-linking studies of ClpB mutants and deletion variants
confirmed the findings obtained by gel filtration analysis; how-
ever, qualitative differences with respect to the oligomerization
deficiencies were revealed. Isolated AAA-domains (1–409 and
551–857) stayed as monomers in presence of glutaraldehyde,
whereas in case of the ClpB ⌬SSD variant (aa 1–758) and a
longer version of the first AAA module (aa 1–567) some dimeric
and trimeric species were observed (Fig. 3B). Mutating the
Walker A motif (K212A) or the Arg finger (R332A) of the first
AAA-domain resulted in the formation of mixtures of oligo-
meric species, ranging from monomers to tetramers in case of
K212A and monomers to hexamers for R332A (Fig. 3Aand
Table I). Interestingly, the observed oligomerization defects
were nucleotide-independent, indicating that the mutated res-
idues are also important for subunit interactions within the
oligomer in the absence of ATP. The involvement of these
charged residues in ClpB assembly can also explain the ob-
served salt sensitivity of ClpB and Hsp104 oligomerization (24,
25). All other ClpB variants exhibited cross-linking character-
istics indistinguishable from wild type ClpB with exception of
ClpB-(⌬410 –532) that misses the linker region. This variant
exhibited only a ladder of cross-linking products in the absence
of nucleotide and required ATP for full oligomerization. The
observed assembly defect was, however, not primarily caused
by the absence of the linker region but rather by the introduc-
tion of additional amino acids at each boundary of the linker
segment, resulting from the construction strategy of this dele-
tion variant (see “Experimental Procedures”). Thus a control
construct (termed ClpB ⌬LBL), carrying the same additional
amino acids and the reinserted linker region, also exhibited
the same oligomerization defects as ClpB-(⌬410 –532) (data
FIG.1.Proposed domain organization of E. coli ClpB and conserved motifs, involved in ATP binding and ATP hydrolysis. ClpB
consists of an N-terminal domain, two nucleotide binding domains (ATP-1 and ATP-2), which are separated by a linker region, and a C-terminal
SSD. Both nucleotide binding domains contain the Walker A (
206
GX
4
GKT
213
and
605
GX
4
GKT
612
) and Walker B (
276
Hy
2
DE
279
and
675
Hy
2
DE
678
)
motifs, where X⫽any amino (aa) and Hy ⫽hydrophobic amino acids. Conserved Arg residues (R) present in both nucleotide binding domains
(Arg-332 and Arg-756) are proposed to serve as Arg fingers, contacting the ATP bound to an adjacent subunit. The invariant sensor 2 motif
(
813
GAR
815
) of the SSD of Hsp100 proteins potentially sense the nucleotide status of ATP-2. Conserved residues of these motifs (marked in
boldface) were subjected to alanine mutagenesis, as indicated by arrows.
Structure-Function Analysis of ClpB 17617
by guest on December 27, 2015http://www.jbc.org/Downloaded from
not shown), showing that the linker is not essential for
oligomerization.
Finally in a complementary approach we examined the in-
trinsic fluorescence of ClpB as a potential tool to monitor the
oligomeric state. ClpB contains 2 tryptophan residues (Trp-462
and Trp-543). We determined whether tryptophan fluorescence
changes in a nucleotide-dependent fashion. In the absence of
nucleotide, ClpB exhibited a fluorescence emission maximum
at 350.5 nm. Addition of ATP or ADP caused a blue shift in the
fluorescence maximum to 346.5 nm, and a stronger decline of
fluorescence intensity beyond the maximum was observed (Fig.
4A). This change was more pronounced in the presence of ATP
compared with ADP. Several findings suggest that the ob-
served changes in fluorescence are caused by oligomerization of
ClpB. First, the blue shift was also observed when ClpB was
incubated in low salt buffer in the absence of nucleotide (Fig. 4).
Such buffer conditions have been shown for several ClpB ho-
mologues to stabilize the oligomer in the absence of nucleotide
(25, 26). Consistently, the ATP-dependent blue shift was re-
verted under high ionic strength conditions (addition of 100 mM
(NH
4
)
2
SO
4
or 300 mMNH
4
Cl), conditions which have been
shown to inhibit ClpB activity (10). De-oligomerization of ClpB
in the additional presence of 100 mM(NH
4
)SO
4
or 300 mM
NH
4
Cl was also observed in cross-linking experiments. Such
high salt conditions resulted in a strong reduction of ClpB
cross-linking efficiency in the absence or presence of ATP (Fig.
4Band data not shown).
Second, the observed changes in fluorescence also became
nucleotide-independent with increasing ClpB concentrations; a
complete blue-shift was revealed in presence of 4
MClpB (data
not shown). Finally, the analysis of ClpB mutants and deletion
variants revealed a direct correlation between the ability to
form hexamers and the observed changes in tryptophan fluo-
rescence; the ATP-dependent blue shift was not or was only
partially observed for ClpB variants with oligomerization de-
fects (Table I).
In order to dissect the individual contributions of the two
tryptophan residues to the changes in the fluorescence spec-
trum, a ClpB variant (W543F) with only a single tryptophan
residue (Trp-462) was constructed. Although this variant was
not affected in oligomerization and exhibited full chaperone
activity in vitro, nucleotide-dependent changes in ClpB fluores-
cence were no longer observed (data not shown). We conclude
that conformational changes in the close vicinity of Trp-543,
driven by oligomerization, must be responsible for the observed
blue shift in tryptophan fluorescence of ClpB.
Nucleotide-dependent Conformational Changes within ClpB
Revealed by Partial Proteolysis—The ability of Hsp100 mu-
tants to form oligomers is commonly used as a criteria for their
structural integrity. We additionally looked for conformational
changes of wild type ClpB in response to nucleotides, and we
checked whether such structural rearrangements were pre-
served in the constructed ClpB variants. ClpB was subjected to
limited proteolysis by thermolysin or subtilisin in the absence
or presence of nucleotides. In the absence of nucleotides two
highly stable fragments were recovered (Fig. 5A). Fragments
were N-terminally sequenced and identified by mass spectrom-
etry (Fig. 5B). The larger degradation products (aa 3–331 and
3–351) corresponded to the N-terminal domain of ClpB and the
core domain of the first AAA module. The second fragment (aa
536 –756) represented the core domain of the second AAA mod-
ule. Addition of ATP or ADP slowed the proteolysis consider-
ably and resulted in stabilization of full-length ClpB and the
occurrence of two other stable cleavage products. The first
fragment (aa 3–331 and 3–351) was already obtained by cleav-
age in the absence of nucleotide, although the second fragment
(aa 537–857 and 551–857) corresponded now to the complete
second AAA module. In the presence of ATP
␥
S cleavage was
further limited, and reduced amounts of the smaller fragments
were obtained (Fig. 5A).
Partial proteolysis of ClpB variants revealed that nucleotide-
dependent stabilization of wild type ClpB can be attributed to
hexamer formation. Mutants with oligomerization deficiencies
(K212A, ClpB ⌬SSD) or reduced stability of hexamers (ClpB-
(⌬410 –532)) were degraded more quickly in the presence of
ATP compared with WT ClpB and thus did not exhibit stabili-
zation of the full-length version (Fig. 5C). The observed defects
of ClpB-(⌬410 –532), missing the linker region, in ATP-depend-
ent stabilization could again be attributed to the presence of
the two additional amino acids at the domain boundary be-
cause the control construct (ClpB ⌬LBL) with the reinserted
linker region exhibited the same reduced proteolytic stability.
Isolated AAA-domains (1–409 and 551–857) did not react to
ATP addition and were processed rapidly to stable cleavage
products (similar to the fragments obtained for WT ClpB in the
FIG.2.Oligomerization of ClpB wild type and mutant deriva-
tives analyzed by gel filtration chromatography. A, elution pro-
files of ClpB wild type (WT) were recorded in the absence (⫺)or
presence of nucleotides (2 mMATP/ADP) in the running buffer (50 mM
Tris, pH 7.5, 20 mMMgCl
2
, 150 mMKCl, 10% (v/v) glycerol). Elution
positions of protein standards (thyroglobulin (669 kDa), ferritin (440
kDa), catalase (232 kDa), and IgG (158 kDa)) are given. B, elution
profiles of the indicated ClpB mutants were recorded in the presence of
ATP (2 mM) in the running buffer.
Structure-Function Analysis of ClpB17618
by guest on December 27, 2015http://www.jbc.org/Downloaded from
absence of nucleotides). In contrast, a longer version of the first
AAA-domain (1–567) in the presence of ATP exhibited some
protection from degradation. Further structural analysis of
ClpB point mutants surprisingly revealed that conformational
changes within each AAA-domain can occur independently of
structural deficiencies in the other AAA module. Thus, the
Walker A mutant K212A of the first AAA-domain, although
deficient in oligomerization, still exhibited ATP-dependent con-
formational changes in the second AAA-domain (stabilization
of the second AAA module). On the other hand structural
changes within the second AAA-domain were not or were only
partially observed in the case of K611A and
813
AAA
815
, ClpB
mutants that nevertheless showed stabilization of the full-
length protein (Fig. 5C). Thus oligomerization protects full-
length ClpB even in case of mutants that do not bind nucleotide
tightly at the second AAA module (K611A;
813
AAA
815
). We
conclude that stabilization of full-length ClpB against proteo-
lytic degradation primarily reflects binding of ATP to the first
AAA-domain. These data also demonstrate that the ability to
form oligomers is not per se a sufficient criteria for probing the
structural integrity of the ClpB mutants. Structural integrity
with respect to oligomerization and stability during partial
proteolysis was only completely preserved in Walker B mu-
tants of ClpB, although Walker A and the sensor 2 mutants
exhibited significant structural deficiencies. Interestingly, the
double Walker B mutant (E279A/E678A) was much more re-
TABLE I
Oligomerization of ClpB fragments and mutants
Assembly of the indicated ClpB derivatives was analyzed in the presence of 2 mMATP by gel filtration chromatography, cross-linking, and
tryptophan fluorescence. Size of ClpB oligomers were calculated from a calibration curve with protein standards. Cross-link products formed after
10 min of incubation with glutaraldehyde were analyzed by SDS 4 –12% PAGE. Most populated species are underlined. ATP-dependent
oligomerization of ClpB is coupled to a blue shift in the maximum of fluorescence (346.5 nm instead of 350.5 nm in the absence of nucleotide). The
maximum of ClpB fluorescence is indicated. ND, not determined.
Gel filtration
chromatography Cross-linking products Maximum of ClpB
fluorescence
ClpB fragments
1–857 (wild type) Hexamer Hexamers 346.5
143–857 (⌬N) Hexamer Hexamers 346.5
1–758 (⌬SSD) Monomer Monomers-trimers 351
⌬410–532 (⌬linker) Hexamer Hexamers 348.5
1–409 Monomer Monomer ND
1–567 Monomer Monomer ND
551–857 Monomer Monomers-dimers ND
ClpB mutants
K212A (Walker A) Dimer Monomers-tetramers 351
K611A (Walker A) Hexamer Hexamers 347
K212A/K611A Dimer Monomers-tetramers 351
E279A (Walker B) Hexamer Hexamers 346.5
E678A (Walker B) Hexamer Hexamers 346.5
E279A/E678A Hexamer Hexamers 346.5
R332A (Arg finger) Dimer Monomer-hexamers 348.5
R756A (Arg finger) Hexamer Hexamers 346.5
813AAA815 (sensor 2) Hexamer Hexamers 346
FIG.3.Assembly of ClpB wild type
and derivatives revealed by cross-
linking. A,1
MClpB wild type (WT) and
the indicated point mutants were incu-
bated for 5 min at 30 °C in the absence of
nucleotides (lane 2) or the presence of 2
mMATP (lane 3) or ADP (lane 4). Cross-
linking reactions were initiated by addi-
tion of glutaraldehyde and proceeded for
10 min. Cross-linked proteins were sepa-
rated on SDS (4 –12%)-polyacrylamide
gels, followed by silver staining. Incuba-
tion of ClpB proteins in the absence of
cross-linker (lane 1) served as control. B,
ClpB deletion variants were incubated in
the absence (lane 2) or presence of 2 mM
ATP (lane 3) for 5 min at 30 °C. Cross-
linking was performed as described
above. Incubation of truncated ClpB spe-
cies in the absence of cross-linker (lane 1)
served as control.
Structure-Function Analysis of ClpB 17619
by guest on December 27, 2015http://www.jbc.org/Downloaded from
sistant toward proteolysis than the single Walker B mutants
and ClpB wild type. Because this mutant is deficient in ATP
hydrolysis (see below), the occurrence of the other fragments
(aa 3–331 and 3–351 and 537–857 and 551–857) in ClpB wild
type (and various ClpB mutants) can be attributed to ATP
hydrolysis occurring during the digestion reaction. These data
also indicate that ClpB therefore adopts a different conforma-
tion if both AAA-domains have ATP bound, as compared with
the situation where only one AAA-domain has ATP bound and
the other domain has ADP bound.
ATPase Activities of ClpB Mutants and Deletion Variants—
ATPase activities of ClpB variants were determined in the
absence or presence of the artificial substrate casein, which is
known to stimulate ATP hydrolysis by ClpB (27). Wild type
ClpB exhibited an ATPase rate of 0.021/s, and the ATPase
activity was increased by 3–4-fold in the presence of saturating
concentrations of
␣
-casein (20-fold excess over ClpB mono-
mers). ClpB ⌬N had a slightly increased ATPase activity but
was less stimulable by casein, in agreement with published
data (23, 28). Removal of the linker region (ClpB-(⌬410 –532))
strongly decreased the basal ATPase activity by 4-fold, al-
though stimulation by casein was not affected. Similar results
were obtained with the control construct ClpB ⌬LBL, bearing a
reinserted linker, indicating that the reduced basal ATPase
activity is primarily caused by the observed oligomerization
deficiencies. All other deletion variants (aa 1–409, 1–567, 551–
857, and 1–758) with even stronger defects in oligomerization
did not exhibit any ATPase activity, even in presence of casein,
indicating that ATP binding and/or hydrolysis is strictly linked
to the formation of hexamers. Consistent with this hypothesis,
the basal ATPase activities of mutants with oligomerization
defects were sensitive to the buffer conditions; ATP activities of
K212A and Arg-332 were enhanced 2–3-fold in low salt buffer,
conditions that favor oligomerization (data not shown). In
agreement with these findings, addition of 100 mM(NH
4
)
2
SO
4
or 300 mMNH
4
Cl diminished the ATPase rate of full-length
ClpB or ClpB ⌬N 3- and 5-fold, respectively (Table II, data not
shown). Further analysis of Walker A and Walker B single
mutants, bearing only one functional AAA-domain, revealed
that especially the first AAA-domain was sensitive toward high
FIG.4.A, nucleotide-dependent blue shift in the wavelength of max-
imum ClpB fluorescence. Emission spectra of tryptophan fluorescence
of ClpB (0.5
M) in the absence of nucleotide (black line) or the presence
of2m
MATP (gray line) or ADP (thin black line) in buffer A (50 mMTris,
pH 7.5, 150 mMKCl, 20 mMMgCl
2
,2mMDTT) are shown. Additionally,
ClpB fluorescence was recorded in low salt buffer (buffer A without 150
mMKCl) (black dotted line) or in the presence of 2 mMATP in high salt
buffer (buffer A plus 100 mM(NH
4
)
2
SO
4
)(gray dotted line). The maxi-
mum values in the wavelength of tryptophan fluorescence are indi-
cated. B, high salt concentrations inhibit cross-linking of ClpB to oligo-
meric species. Cross-linking of ClpB (1
M) by glutaraldehyde was
performed in the absence or presence of ATP (2 mM) and/or (NH
4
)
2
SO
4
(100 mM). Cross-linked proteins were separated on SDS (4 –12%)-poly-
acrylamide gels, followed by silver staining.
FIG.5. Nucleotide-dependent structural changes in ClpB re-
vealed by limited proteolysis. A, partial proteolysis of ClpB wild
type by thermolysin and subtilisin was performed in the absence of
nucleotide or in the presence of 2 mMATP/ADP/ATP
␥
S for 60 min.
Cleavage products were separated on SDS (15%)-polyacrylamide gels,
followed by silver staining. Cleavage sites were identified by N-terminal
sequencing and mass spectrometry. B, proposed domain organization of
ClpB. Domains were defined from sequence analysis, secondary struc-
ture prediction, and results obtained from partial proteolysis. ClpB
domains are as follows: N-domain (1–144), ATP-1 (144 –340), C-1 (341–
409 and 525–550), the linker region (410 –525), ATP-2 (550–755), and
C-2 (755–857). ATP-1/2 corresponds to the core domains of the AAA
modules. C-1/2 represents the C-terminal helical domains of the AAA
modules. Sites of cleavage by thermolysin and subtilisin are indicated
(open arrows, cleavage site in absence of ATP; filled arrows, cleavage
sites in presence of nucleotides). C, partial proteolysis of ClpB mutants
and fragments by thermolysin. Proteins were incubated in the absence
or presence of ATP for 60 min. Cleavage products were separated on
SDS (15%)-polyacrylamide gels, followed by silver staining.
Structure-Function Analysis of ClpB17620
by guest on December 27, 2015http://www.jbc.org/Downloaded from
ionic strength conditions. Addition of 100 mM(NH
4
)
2
SO
4
strongly reduced the ATPase activity of mutants, in which the
second AAA-domain was inactivated (K611A and E678A),
whereas variants with only an active second AAA module
(E279A and K211A) were only partially affected (Table II),
thereby underlining the functional importance of the first AAA-
domain for ClpB assembly. Higher concentrations of ammo-
nium sulfate (200 mM) completely inhibited ATP hydrolysis by
ClpB (data not shown).
Single Walker A or Walker B mutations resulted in a com-
plete loss of ATPase activity in the corresponding AAA-domain,
because double Walker A or Walker B mutants were deficient
in ATP hydrolysis, even in presence of casein. Because signif-
icant ATPase activities were measurable for all Walker A and
Walker B single mutants, both ATPase domains seemed to
contribute to the basal ATPase activity of ClpB. Additionally,
both AAA modules were stimulated to similar degrees by ca-
sein. Variations in the basal ATPase activities of ClpB point
mutants compared with wild type ClpB indicate a communica-
tion between both AAA-domains. The basal ATPase rate of
ClpB E678A was increased 4-fold compared with wild type.
However, analysis of other ClpB mutants revealed that no easy
conclusions with respect to the directionality of the communi-
cation between the AAA-domains can be drawn. Mutating the
Arg finger of the second AAA module (R756A) resulted in a
nearly complete loss of ATPase activity, although the overall
structural integrity was preserved as for the single Walker B
mutant (E678A). The signaling between both AAA-domains
seemed to be rather complex, and subtle conformational
changes within ClpB mutants can obviously have completely
different consequences on the other AAA module.
The mechanism of ATPase stimulation by casein is un-
known. Because ATP hydrolysis by ClpB is sensitive to its
oligomeric status, we tested the possibility that interaction of
ClpB with casein stabilizes the hexameric state, thereby in-
creasing the ATPase activity. We compared the ATPase activ-
ity of ClpB concentration at different protein concentrations
both in the absence and presence of casein. In the absence of
substrate, low ClpB concentrations (0.1–0.2
M) had no activity
or a strongly reduced specific ATPase activity (Fig. 6). Increas-
ing ClpB concentrations activated ATP hydrolysis in a cooper-
ative manner as revealed by the sigmoidal shape of the curve.
In contrast, in the presence of casein high ATPase activities
were already determined at low ClpB concentrations (0.1
M),
implying that stabilization of ClpB hexamers by substrates
causes induction of the ATPase activity. Such stabilization by
casein was indeed observed by gel filtration analysis for the
ClpB Walker B mutant E279A/E678A, which is deficient in
ATP hydrolysis.
2
Besides stabilization of the oligomeric state,
conformational changes within ClpB, triggered by substrate
binding, may also contribute to the stimulation of ATP hydrol-
ysis, because saturating ClpB concentrations (1
M) still
showed a lower specific ATPase activity in the absence of casein
(Fig. 6).
Chaperone Activities of ClpB Derivatives—We tested the
chaperone activities of the various ClpB variants in vitro by
following the ClpB/DnaK-dependent reactivation of heat-ag-
gregated MDH and firefly luciferase. Resolubilization of MDH
aggregates by ClpB/DnaK was directly followed by measuring
the decrease of aggregate turbidity in light scattering experi-
ments. Disaggregation of aggregated luciferase was followed by
determining the refolding rate of luciferase in the presence of
the bi-chaperone system. All disaggregation reactions were
performed in the presence of non-saturating ClpB concentra-
tions which allowed for sensitive detection of potential activity
defects. Results are summarized in Table III.
All ClpB deletion variants (aa 1–409, 1–567, 551–857, and
1–758) with severe oligomerization defects were inactive in
MDH and luciferase disaggregation. Deletion of the linker re-
gion (ClpB-(⌬410 –532)) also caused a complete loss of the dis-
aggregation activity of ClpB. Importantly, this inactivation was
not caused by the observed oligomerization deficiency of ClpB-
(⌬410 –532), because the control construct ClpB ⌬LBL with the
reinserted linker region exhibited significant disaggregation
activity (60% of ClpB wild type; data not shown). ClpB ⌬N had
the same chaperone activity as full-length ClpB. Full disaggre-
gation activity of ClpB ⌬N was also observed when the ClpB
concentrations in the activity assays were further reduced,
thereby increasing the dependence of the disaggregation on
ClpB (data not shown).
ClpB variants with point mutations in conserved motifs ex-
hibited none or only partial chaperone activity (Table III). In
general a disaggregation activity was only observed for mu-
tants that are still able to form oligomers. Thus the Walker A
(K212A) and Arg finger (R332A) mutations of the first AAA-
domain, which are deficient in hexamer formation, were inac-
2
J. Weibezahn, C. Schlieker, B. Bukau, and A. Mogk, manuscript
in preparation.
TABLE II
ATPase activities of ClpB fragments and mutants
ATPase activities of the indicated ClpB derivatives were determined
in buffer A. The factors of ATPase stimulation in the presence of 0.25
mg/ml
␣
-casein and the factors of ATPase inhibition in the presence of
100 mMammonium sulfate are given. ND, not determined.
ATPase
activity
(1/s)
Factor of
ATPase
stimulation
by
␣
-casein
Factor of
ATPase
inhibition
by sulfate
ClpB fragments
1–857 (wild type) 0.021 3.3 3
143–857 (⌬N) 0.037 1.9 4.9
1–758 (⌬SSD) 0 0 ND
⌬410–532 (⌬linker) 0.005 6 ND
1–409 0 0 ND
1–567 0 0 ND
551–857 0 0 ND
ClpB mutants
K212A (Walker A) 0.006 4.4 1.4
K611A (Walker A) 0.019 5.3 4.2
K212A/K611A 0 0 ND
E279A (Walker B) 0.023 3.8 1.7
E678A (Walker B) 0.089 2.7 5.3
E279A/E678A 0 0 ND
R332A (Arg finger) 0.019 2.4 ND
R756A (Arg finger) 0.003 0 ND
813AAA815 (SSD motif) 0.036 2.8 ND
FIG.6. The ATPase activity of ClpB depends on protein con-
centration and protein substrates. The specific ATPase activity of
ClpB was measured at different protein concentrations in the absence
or presence of
␣
-casein (0.25 mg/ml).
Structure-Function Analysis of ClpB 17621
by guest on December 27, 2015http://www.jbc.org/Downloaded from
tive. In contrast, the corresponding mutations in the second
AAA-domain (K611A and R756A) exhibited low chaperone ac-
tivities to varying degrees (below 10% of the WT control).
Partial activities were also obtained in case of single Walker B
mutants, although the double mutant (E279A/E678A) did not
have any disaggregation activity. These findings indicate that
whereas ATP hydrolysis is strictly required for chaperone ac-
tivity, a lack of ATP hydrolysis in one AAA-domain does not
completely inhibit the disaggregation activity.
Disaggregation activities of all ClpB variants were also
tested in vivo by complementation of the phenotypes of ⌬clpB
mutant cells by IPTG induction of plasmid-encoded clpB mu-
tant alleles. Resolubilization of protein aggregates, formed dur-
ing severe heat stress to 45 °CinE. coli cells, is severely
affected in ⌬clpB mutants. This deficiency is directly linked to
a strongly reduced survival rate of ⌬clpB mutant cells at lethal
(50 °C) temperatures compared with wild type cells (thermo-
tolerance). We therefore tested the ClpB variants for their
ability to re-establish protein disaggregation and thermotoler-
ance in ⌬clpB mutant cells (Fig. 7). Complementation studies
were performed in the presence of 25
MIPTG, leading to
3–4-fold increased ClpB levels as compared with heat-shocked
wild type cells lacking any plasmid (data not shown). In sum-
mary the in vivo disaggregation activities reflected those ob-
served in vitro. Only wild type ClpB and the ClpB ⌬N could
efficiently mediate the resolubilization of protein aggregates
and the development of thermotolerance (Fig. 7 and Table III).
Constructs with partial chaperone activities in vitro, such as
E279A, also exhibited some activity in vivo.
DISCUSSION
Here we present a detailed structure-function analysis of the
AAA⫹chaperone ClpB, which mediates resolubilization of pro-
tein aggregates in cooperation with the DnaK chaperone sys-
tem. Oligomerization of ClpB was dependent on both AAA-
domains; however, each domain seemed to play a different role
in ClpB assembly. Deletion of the
␣
-helical C-domain of the
second AAA-domain (ClpB ⌬SSD) resulted in a severe defect in
oligomerization defect. ClpB ⌬SSD stayed predominantly mon-
omeric even in cross-linking experiments. On the other hand
we could demonstrate that ATP binding to the first AAA-
domain was also necessary for stabilizing the ClpB hexamer.
Several observations support this model. First, the Walker A
(K212A) and Arg finger (R332A) mutants of the first AAA
module did not form stable hexamers in the presence of ATP,
whereas corresponding mutations in the second nucleotide
binding domain (K611A and R756A) had no influence on ClpB
oligomerization, although these mutants exhibited severe
structural deficiencies. Second, the introduction of amino acids
into the C-domain of the first AAA module (ClpB-(⌬410 –532)
and ClpB ⌬LBL) also resulted in destabilization of ClpB oli-
gomers, underlining the importance of this domain in hexa-
merization. We also could show that high salt conditions (ad-
dition of 100 mM(NH
4
)
2
SO
4
) cause dissociation of ClpB
oligomers by interfering predominantly with subunit contacts
between the first AAA-domains. Interestingly, the recently de-
termined structure of monomeric ClpA suggests that electro-
static interactions between the first AAA-domains of ClpA
could play a much more important role than the second AAA
modules in protein oligomerization (29). These findings might
also explain the observed salt sensitivity of ClpB oligomeriza-
tion and underline the functional importance of the first AAA-
domain in this process. Finally, the observed changes in fluo-
rescence of Trp-543, located in the C-domain of the first AAA
module, revealed a conformational rearrangement of this re-
gion in response to nucleotides. The observed blue shift (4 nm)
TABLE III
Chaperone activities of ClpB mutants and fragments
Chaperone activity (disaggregation rate of aggregated MDH; refold-
ing rate of aggregated luciferase) of ClpB wild type was set as 100%. In
vivo activity reflects the ability of ClpB derivatives to restore thermo-
tolerance and protein disaggregation in E. coli ⌬clpB mutant cells (⫹,
full activity; (⫺), residual activity; ⫺, no activity). ND, not determined.
Disaggregation
of aggregated
MDH
Refolding of
aggregated
luciferase
In vivo
activity
% activity
ClpB fragments
1–857 (wild type) 100 100 ⫹
143–857 (⌬N) 98 99 ⫹
1–758 (⌬SSD) 0 0 ⫺
⌬410–532 (⌬linker) 0 0 ⫺
1–409 0 0 ND
1–567 0 0 ND
551–857 0 0 ND
ClpB mutants
K212A (Walker A) 0 0 ⫺
K611A (Walker A) 4–58–10 (⫺)
K212A/K611A 0 0 ⫺
E279A (Walker B) 7–98–10 (⫺)
E678A (Walker B) 2–34–5(⫺)
E279A/E678A 0 0 ⫺
R332A (Arg finger) 0 0 ⫺
R756A (Arg finger) 0.5–1 0.5–1⫺
813AAA815 (SSD motif) 0 0 ⫺
FIG.7. In vivo activity of ClpB derivatives. A,E. coli wild type
(MC4100), ⌬clpB mutant cells, and ⌬clpB mutants, bearing plasmid-
encoded clpB alleles under the control of an IPTG-regulatable promo-
tor, were grown in LB medium at 30 °C in the presence of 25
MIPTG
to mid-exponential phase. Subsequently strains were heat-shocked to
50 °C and incubated for the indicated time. Various dilutions of the
stressed cells were spotted on LB plates and incubated at 30 °C. After
24 h, colony numbers were counted, and survival rates were calculated
in relation to unstressed cells. B, cultures of ⌬clpB mutant strains,
bearing plasmid-encoded clpB alleles under the control of an IPTG-
regulatable promotor, were grown in LB medium in the presence of 25
MIPTG at 30 °C to logarithmic phase. Cells were then shifted to 45 °C
for 30 min, followed by a recovery phase at 30 °C for 60 min. Aggregated
proteins were isolated before (0 min) and after the recovery phase (60
min) and analyzed by SDS-PAGE followed by staining with Coomassie
Brilliant Blue.
Structure-Function Analysis of ClpB17622
by guest on December 27, 2015http://www.jbc.org/Downloaded from
in the wavelength for maximum fluorescence could be attrib-
uted to ClpB oligomerization. We suggest that changes in Trp-
543 fluorescence reflect interactions of the first C-domain not
only with itself but also adjacent ATPase domains, thereby
leading to increased shielding of Trp-543 and stabilization of
the oligomer. Nucleotide-dependent conformational changes of
the C-terminal
␣
-helical domain was also demonstrated by
limited proteolysis. Whereas the C-domain was rapidly de-
graded in the absence of nucleotides, it became largely resist-
ant to proteases upon nucleotide addition (Fig. 5B). Similarly
the C-domain of the second AAA module was also protected by
addition of nucleotides. This protection is likely to be caused by
the interaction of C-domains with their own AAA-domain and
that of adjacent subunits, thereby becoming less accessible to
proteases. Consistently, ClpB mutants with defects in nucleo-
tide binding (K212A and K611A;
813
AAA
815
) did not exhibit
stabilization of the corresponding C-domains.
A functional importance of the first ATPase domain for
Hsp100 oligomerization has also been reported for E. coli ClpA
(30, 31) and ClpB from Thermus thermophilus (32). It is in-
triguing that the contributions of the individual AAA-domains
differ in the ClpB homologues Hsp104 and Hsp78 from Saccha-
romyces cerevisiae. Here mutations of the Lys residue in the
Walker A motif of the second AAA-domain resulted in a loss of
ability to form hexamers (33–35).
Hexamerization of ClpB is a prerequisite for its chaperone
activity. Because ATP is bound at the interface of two neigh-
boring subunits in the hexamer, oligomerization additionally
influences the ATPase activity of ClpB. Consequently high salt
conditions inhibit ATP hydrolysis by ClpB. Furthermore, ClpB
variants with defects in oligomerization exhibited little or no
ATPase activity. On the other hand, the stimulation of ATP
hydrolysis by substrates, such as
␣
-casein, is most likely be-
cause of the stabilization of ClpB oligomers.
Mutations of conserved Lys and Glu residues in the Walker
A and Walker B motifs of both AAA modules completely abol-
ished ATP hydrolysis by the corresponding AAA-domain. Be-
cause single mutants still retained significant ATPase activity,
both AAA-domains seem to contribute to the basal rate of ATP
hydrolysis by ClpB. Interestingly, ClpB homologues from dif-
ferent organisms differ significantly in their ATPase cycle.
Although both AAA-domains of ClpB from T. thermophilus (26,
32) also contribute to the basal ATPase activity, ATP hydroly-
sis by the yeast homologue Hsp104 is dominated by the first
AAA module (25). Such differences may potentially explain the
observed species specificity in the cooperation of ClpB/Hsp104
proteins with the corresponding Hsp70 partners (36). Varia-
tions in the basal ATPase activity of ClpB mutants indicate
communication between both AAA-domains. However, differ-
ent mutations in the same AAA-domain (E678A and R756A)
exhibited different influences on the ATPase activity of the
other AAA module. The signaling between both AAA-domains
appears to be rather complex, and subtle conformational
changes within ClpB mutants can have completely different
consequences on the other AAA module.
The function of the N-domain still remains enigmatic. Iso-
lated N-domains from ClpA and ClpB have been shown to form
stable, monomeric domains, which are probably separated from
the AAA-domains by a flexible linker (7, 37, 38). Consistent
with this model, N-domains are not essential for oligomeriza-
tion of ClpA or ClpB (23, 37, and this work). The connector is
apparently sterically inaccessible to proteases because partial
proteolysis of ClpB did not release significant amounts of iso-
lated N-domains but rather produced a fragment comprising
the N-domain and the core domain of the first AAA module (aa
1–353). Similar findings have been reported for ClpA (37). The
reported consequences of N-terminal deletions of ClpB on its
chaperone activity are contradictory. Zolkiewski and co-work-
ers (23) showed that a ClpB variant, starting from the internal
start site (aa 149 –857), was completely inactive in refolding of
aggregated luciferase. In contrast, a ClpB ⌬N-(143–857) vari-
ant showed the same chaperone activity as wild type ClpB both
in vitro and in vivo (resolubilization of protein aggregates and
development of thermotolerance). In agreement with these
data, an N-terminal truncated ClpB derivative of Synechococ-
cus sp. PCC7942 conferred the same degree of thermotolerance
in vivo as full-length ClpB; likewise, the N-terminal truncated
version of T. thermophilus ClpB was also shown to be active in
protein disaggregation in vitro (39, 40). The existence of ClpB
homologues completely lacking an N-domain in Mycoplasma
sp. also argues against an essential function of the N-domains
in the disaggregating activities of ClpB (41, 42).
What might be the function of the N-domain with respect to
these conflicting results? N-domains have been proposed to
mediate substrate binding; however, the reported activities of
N-terminally truncated ClpB variants clearly rule out an es-
sential function in this process. Interestingly, defects in sub-
strate binding were also not the basis of the inactivation of
ClpB-(149 –857), because this deletion variant interacted with
casein and unfolded luciferase indistinguishable from ClpB
wild type (38). N-domains may therefore be involved in a mech-
anism for coupling the ATPase cycle of ClpB with its proposed
unfolding activity. Because a longer deletion of the N-domain,
including parts of the flexible linker to the first AAA-domain,
has been reported to be inactive, this linker could potentially
contribute to induced structural changes in bound substrates.
Alternatively, N-domains may be involved in other ClpB activ-
ities, which are so far unknown and are not related to protein
disaggregation. Such new activities have been described for a
second ClpB homologue in Synechococcus, which is essential for
cell viability but is not involved in thermotolerance (43). Inter-
estingly, the N-domain of this ClpB variant seems to be crucial
for its unknown activity and may serve as binding sites for
special substrates or, alternatively, for specific adaptor pro-
teins. Binding of adaptor proteins to N-domains has been dem-
onstrated for ClpA (44) and the AAA protein p97 (45, 46).
The linker region of ClpB was originally suggested to sepa-
rate both AAA-domains (1). We propose that the linker instead
interrupts the C-domain of the first AAA-domain as was sug-
gested recently (47) for the yeast homologue Hsp104. A com-
parison of different ClpB fragments (aa 1–409 and 1–567) with
respect to their oligomerization and their resistance to prote-
olysis supports this model. First, cross-linking studies revealed
that ClpB-(1–567) can form dimeric species in contrast to the
shorter variant (aa 1–409), which remains monomeric under
all conditions. Additionally, partial proteolysis revealed a more
pronounced protection of the full-length version (aa 1–567),
which was not observed for ClpB-(1–409). Formation of dimeric
species and the observed partial stabilization is likely to be
caused by the interaction of the helical C-domain, which can
only be formed in ClpB-(1–567), with its own and an adjacent
AAA-domain. Interestingly, a very similar domain organiza-
tion has been proposed for ClpA (37). In ClpA, which is missing
the linker region, a short basic loop (KRKK) is also inserted
into the C-domain of the first AAA module.
The linker region of ClpB, in contrast to the N-domain, is
essential for chaperone activity. It is proposed to form a 4 times
repeated coiled-coil (48), which is very likely to play an impor-
tant role in ClpB function. Conformational changes in response
to nucleotides within the linker region were shown by its in-
creased proteolytic stability in the context of full-length ClpB.
We assume that the linker is not an integral part of the ClpB
Structure-Function Analysis of ClpB 17623
by guest on December 27, 2015http://www.jbc.org/Downloaded from
hexamer, because oligomerization was still possible in case of a
ClpB variant missing the linker region (ClpB-(⌬410 –532)).
Similarly, the I-domain of HslU, although inserted into the
AAA-domain, forms an independent structural domain and is
exposed at the surface of the oligomer (3, 4). The function of the
linker region is still unknown. The postulated coiled-coil struc-
ture might be involved in protein-protein interaction. Because
the ATPase activity of ClpB-(⌬410 –532) was still strongly
stimulable by casein, the linker region cannot serve as a pri-
mary substrate-binding site, at least for this type of substrate.
Alternatively, the linker region is necessary for coupling ATP
hydrolysis and substrate unfolding, as proposed recently by
Lindquist and co-workers (47).
Acknowledgments—We thank D. Dougan and K. Turgay for discus-
sions and critical reading of the manuscript. We also thank A. Schulze-
Specking for technical assistance.
REFERENCES
1. Schirmer, E. C., Glover, J. R., Singer, M. A., and Lindquist, S. (1996) Trends
Biochem. Sci. 21, 289 –296
2. Neuwald, A. F., Aravind, L., Spouge, J. L., and Koonin, E. V. (1999) Genome
Res. 9, 27–43
3. Bochtler, M., Hartmann, C., Song, H. K., Bourenkov, G. P., Bartunik, H. D.,
and Huber, R. (2000) Nature 403, 800 –805
4. Sousa, M. C., Trame, C. B., Tsuruta, H., Wilbanks, S. M., Reddy, V. S., and
McKay, D. B. (2000) Cell 103, 633–643
5. Li, J., and Sha, B. (2002) J. Mol. Biol. 318, 1127–1137
6. Smith, C. K., Baker, T. A., and Sauer, R. T. (1999) Proc. Natl. Acad. Sci.
U. S. A. 96, 6678 –6682
7. Lo, J. H., Baker, T. A., and Sauer, R. T. (2001) Protein Sci 10, 551–559
8. Banecki, B., Wawrzynow, A., Puzewicz, J., Georgopoulos, C., and Zylicz, M.
(2001) J. Biol. Chem. 276, 18843–18848
9. Kim, Y. I., Levchenko, I., Fraczkowska, K., Woodruff, R. V., Sauer, R. T., and
Baker, T. A. (2001) Nat. Struct. Biol. 8, 230 –233
10. Goloubinoff, P., Mogk, A., Peres Ben Zvi, A., Tomoyasu, T., and Bukau, B.
(1999) Proc. Natl. Acad. Sci. U. S. A. 96, 13732–13737
11. Mogk, A., Tomoyasu, T., Goloubinoff, P., Ru¨diger, S., Ro¨der, D., Langen, H.,
and Bukau, B. (1999) EMBO J. 18, 6934 –6949
12. Motohashi, K., Watanabe, Y., Yohda, M., and Yoshida, M. (1999) Proc. Natl.
Acad. Sci. U. S. A. 96, 7184–7189
13. Zolkiewski, M. (1999) J. Biol. Chem. 274, 28083–28086
14. Laufen, T., Mayer, M. P., Beisel, C., Klostermeier, D., Reinstein, J., and
Bukau, B. (1999) Proc. Natl. Acad. Sci. U. S. A. 96, 5452–5457
15. Shevchenko, A., Wilm, M., Vorm, O., and Mann, M. (1996) Anal. Chem. 68,
850 –858
16. Wilm, M., Shevchenko, A., Houthaeve, T., Breit, S., Schweigerer, L., Fotsis, T.,
and Mann, M. (1996) Nature 379, 466 –469
17. Tomoyasu, T., Mogk, A., Langen, H., Goloubinoff, P., and Bukau, B. (2001)
Mol. Microbiol. 40, 397–413
18. Kim, K. I., Cheong, G. W., Park, S. C., Ha, J. S., Woo, K. M., Choi, S. J., and
Chung, C. H. (2000) J. Mol. Biol. 303, 655–666
19. Ogura, T., and Wilkinson, A. J. (2001) Genes Cells 6, 575–597
20. Nicholls, A., Sharp, K. A., and Honig, B. J. (1991) Proteins 11, 281–296
21. Karata, K., Inagawa, T., Wilkinson, A. J., Tatsuta, T., and Ogura, T. (1999)
J. Biol. Chem. 274, 26225–26232
22. Zolkiewski, M., Kessel, M., Ginsburg, A., and Maurizi, M. R. (1999) Protein Sci.
8, 1899 –1903
23. Barnett, M. E., Zolkiewska, A., and Zolkiewski, M. (2000) J. Biol. Chem. 275,
37565–37571
24. Schmid, F. X. (1991) Curr. Biol. 1, 36 –41
25. Hattendorf, D. A., and Lindquist, S. L. (2002) EMBO J. 21, 12–21
26. Schlee, S., Groemping, Y., Herde, P., Seidel, R., and Reinstein, J. (2001) J. Mol.
Biol. 306, 889 –899
27. Woo, K. M., Kim, K. I., Goldberg, A. L., Ha, D. B., and Chung, C. H. (1992)
J. Biol. Chem. 267, 20429 –20434
28. Park, J. H., Lee, Y. S., Chung, C. H., and Goldberg, A. L. (1988) J. Bacteriol.
170, 921–926
29. Guo, F., Maurizi, M. R., Esser, L., and Xia, D. (2002) J. Biol. Chem. 277,
46743–46752
30. Singh, S. K., and Maurizi, M. R. (1994) J. Biol. Chem. 269, 29537–29545
31. Seol, J. H., Baek, S. H., Kang, M.-S., Ha, D. B., and Chung, C. H. (1995) J. Biol.
Chem. 270, 8087–8092
32. Watanabe, Y. H., Motohashi, K., and Yoshida, M. (2002) J. Biol. Chem. 277,
5804 –5809
33. Parsell, D. A., Kowal, A. S., and Lindquist, S. (1994) J. Biol. Chem. 269,
4480 –4487
34. Schirmer, E. C., Queitsch, C., Kowal, A. S., Parsell, D. A., and Lindquist, S.
(1998) J. Biol. Chem. 273, 15546 –15552
35. Krzewska, J., Konopa, G., and Liberek, K. (2001) J. Mol. Biol. 314, 901–910
36. Krzewska, J., Langer, T., and Liberek, K. (2001) FEBS Lett. 489, 92–96
37. Singh, S. K., Rozycki, J., Ortega, J., Ishikawa, T., Lo, J., Steven, A. C., and
Maurizi, M. R. (2001) J. Biol. Chem. 276, 29420 –29429
38. Tek, V., and Zolkiewski, M. (2002) Protein Sci. 11, 1192–1198
39. Clarke, A. K., and Eriksson, M. J. (2000) J. Bacteriol. 182, 7092–7096
40. Beinker, P., Schlee, S., Groemping, Y., Seidel, R., and Reinstein, J. (2002)
J. Biol. Chem. 277, 47160 –47166
41. Fraser, C. M., Gocayne, J. D., White, O., Adams, M. D., Clayton, R. A.,
Fleischmann, R. D., Bult, C. J., Kerlavage, A. R., Sutton, G., Kelley, J. M.,
et al. (1995) Science 270, 397–403
42. Himmelreich, R., Hilbert, H., Plagens, H., Pirkl, E., Li, B. C., and Herrmann,
R. (1996) Nucleic Acids Res. 24, 4420 –4449
43. Eriksson, M. J., Schelin, J., Miskiewicz, E., and Clarke, A. K. (2001) J.
Bacteriol. 183, 7392–7396
44. Dougan, D. A., Reid, B. G., Horwich, A. L., and Bukau, B. (2002) Mol. Cell 9,
673–683
45. Rouiller, I., Butel, V. M., Latterich, M., Milligan, R. A., and Wilson-Kubalek,
E. M. (2000) Mol. Cell 6, 1485–1490
46. Meyer, H. H., Shorter, J. G., Seemann, J., Pappin, D., and Warren, G. (2000)
EMBO J. 19, 2181–2192
47. Cashikar, A. G., Schirmer, E. C., Hattendorf, D. A., Glover, J. R., Ramakrishnan,
M. S., Ware, D. M., and Lindquist, S. L. (2002) Mol. Cell 9, 751–760
48. Celerin, M., Gilpin, A. A., Schisler, N. J., Ivanov, A. G., Miskiewicz, E., Krol,
M., and Laudenbach, D. E. (1998) J. Bacteriol. 180, 5173–5182
Structure-Function Analysis of ClpB17624
by guest on December 27, 2015http://www.jbc.org/Downloaded from
Bernd Bukau
Strub, Wolfgang Rist, Jimena Weibezahn and
Axel Mogk, Christian Schlieker, Christine
and Chaperone Activity
ClpB in Oligomerization, ATP Hydrolysis,
Conserved Motifs of the AAA+ Chaperone
Roles of Individual Domains and
PROTEIN STRUCTURE AND FOLDING:
doi: 10.1074/jbc.M209686200 originally published online March 6, 2003
2003, 278:17615-17624.J. Biol. Chem.
10.1074/jbc.M209686200Access the most updated version of this article at doi:
.JBC Affinity SitesFind articles, minireviews, Reflections and Classics on similar topics on the
Alerts:
When a correction for this article is posted• When this article is cited•
to choose from all of JBC's e-mail alertsClick here
http://www.jbc.org/content/278/20/17615.full.html#ref-list-1
This article cites 48 references, 27 of which can be accessed free at
by guest on December 27, 2015http://www.jbc.org/Downloaded from
