Structure and Molecular Dynamics Simulation of Archaeal Prefoldin

Transcription

Structure and Molecular Dynamics Simulation of Archaeal Prefoldin
doi:10.1016/j.jmb.2007.12.010
J. Mol. Biol. (2008) 376, 1130–1141
Available online at www.sciencedirect.com
Structure and Molecular Dynamics Simulation of
Archaeal Prefoldin: The Molecular Mechanism
for Binding and Recognition of Nonnative
Substrate Proteins
Akashi Ohtaki 1 †, Hiroshi Kida 2 †, Yusuke Miyata 1 †, Naoki Ide 1 †,
Akihiro Yonezawa 1 , Takatoshi Arakawa 1 , Ryo Iizuka 1,3 ,
Keiichi Noguchi 1 , Akiko Kita 2,4 , Masafumi Odaka 1 , Kunio Miki 2,5
and Masafumi Yohda 1 ⁎
1
Department of Biotechnology
and Life Science, Tokyo
University of Agriculture and
Technology, 2-24-16 Naka-cho,
Koganei, Tokyo 184-8588, Japan
2
Department of Chemistry,
Graduate School of Science,
Kyoto University, Sakyo-ku,
Kyoto 606-8502, Japan
3
Laboratory of Bio-Analytical
Chemistry, Graduate School
of Pharmaceutical Sciences,
The University of Tokyo,
7-3-1 Hongo, Bunkyo-ku,
Tokyo 113-0033, Japan
4
Research Reactor Institute,
Kyoto University, Kumatori,
Osaka 590-0494, Japan
5
RIKEN SPring-8 Center at
Harima Institute, Koto 1-1-1,
Sayo, Hyogo 679-5148, Japan
Prefoldin (PFD) is a heterohexameric molecular chaperone complex in the
eukaryotic cytosol and archaea with a jellyfish-like structure containing six
long coiled-coil tentacles. PFDs capture protein folding intermediates or
unfolded polypeptides and transfer them to group II chaperonins for
facilitated folding. Although detailed studies on the mechanisms for
interaction with unfolded proteins or cooperation with chaperonins of
archaeal PFD have been performed, it is still unclear how PFD captures the
unfolded protein. In this study, we determined the X-ray structure of
Pyrococcus horikoshii OT3 PFD (PhPFD) at 3.0 Å resolution and examined the
molecular mechanism for binding and recognition of nonnative substrate
proteins by molecular dynamics (MD) simulation and mutation analyses.
PhPFD has a jellyfish-like structure with six long coiled-coil tentacles and a
large central cavity. Each subunit has a hydrophobic groove at the distal
region where an unfolded substrate protein is bound. During MD
simulation at 330 K, each coiled coil was highly flexible, enabling it to
widen its central cavity and capture various nonnative proteins. Docking
MD simulation of PhPFD with unfolded insulin showed that the β subunit
is essentially involved in substrate binding and that the α subunit
modulates the shape and width of the central cavity. Analyses of mutant
PhPFDs with amino acid replacement of the hydrophobic residues of the β
subunit in the hydrophobic groove have shown that βIle107 has a critical
role in forming the hydrophobic groove.
© 2007 Elsevier Ltd. All rights reserved.
Received 2 October 2007;
received in revised form
28 November 2007;
accepted 5 December 2007
Available online
8 December 2007
Edited by R. Huber
Keywords: prefoldin; group II chaperonin; chaperone; archaea; molecular
dynamics
*Corresponding author. E-mail address: yohda@cc.tuat.ac.jp.
†These authors equally contributed to this work.
Present address: T. Arakawa, Japan Science and Technology Agency, 1-12 Minamiwatarida-cho, Kawasaki, Kanagawa
210-0855, Japan.
Abbreviations used: PFD, prefoldin; PhPFD, Pyrococcus horikoshii OT3 prefoldin; MtPFD, Methanobacterium
thermoautotrophicum prefoldin; MD, molecular dynamics; CS, citrate synthase; GFP, green fluorescent protein; PDB,
Protein Data Bank.
0022-2836/$ - see front matter © 2007 Elsevier Ltd. All rights reserved.
1131
Structure and MD Simulation of Archaeal Prefoldin
Introduction
Molecular chaperones are ubiquitous proteins that
are required for the correct folding, transport, and
degradation of proteins within the cell.1 Molecular
chaperones function by protecting nonnative polypeptides from forming aggregates or by providing
them with appropriate environments where they can
fold properly. Prefoldin (PFD) stabilizes a nonnative
protein and subsequently delivers it to a group II
chaperonin to facilitate the proper folding.2–5 Eukaryotic PFD participates in the maturation of actin and
members of the tubulin family. It captures actin or
tubulin in the unfolded state and transfers it to the
chaperonin-containing TCP-1, the eukaryotic group
II chaperonin, for functional folding. Although there
is neither actin nor tubulin in archaea, both PFD and
group II chaperonins have invariably been identified
in all archaeal species. Eukaryotic PFD is a multiplesubunit complex composed of six polypeptides (two
α-type and four β-type subunits) in the molecular
mass range of 14–23 kDa.2 On the other hand,
archaeal PFD is composed of two kinds of subunits,
two α and four β subunits.
Among archaeal PFDs, Methanobacterium thermoautotrophicum PFD (MtPFD) and Pyrococcus horikoshii
Fig. 1 (legend on next page)
1132
Structure and MD Simulation of Archaeal Prefoldin
Fig. 1. Crystal structure of Pyrococcus PFD. (a) Overall structure of the α2β4 hexameric complex of PhPFD showing side
and bottom views as illustrated by the program MOLSCRIPT.14,15 The crystal structure of PhPFD is composed of one α
subunit and two β subunits in an asymmetrical unit. The biologically active α2β4 hexamer is derived by applying
crystallographic symmetry. The α subunit and two β subunits are shown in red, green (β1 subunit), and blue (β2 subunit).
The symmetrical molecules are shown in light colors. (b) The overall structure of each subunit [the coloring is as in (a)]. The
hydrophobic residues (αLeu11, αLeu14, αVal131, and αVal138 in the α subunit; βVal8, βLeu12, βLeu15, βLeu103, and
βIle107 in the β1 subunit; and βLeu12, βLeu15, βLeu103, and βIle107 in the β2 subunit) at the distal region of each subunit
are in CPK (Corey–Pauling–Koltun) representation. (c) The α- and β-subunit hydrophobic groove structure at the distal
region of the coiled coil. The hydrophobic residues (αLeu11, αLeu14, αVal131, αVal138, βVal8, βLeu12, βLeu15, βLeu103,
and βIle107) are shown by ball-and-stick models.
OT3 PFD (PhPFD) have been characterized in
detail. 6–10 The crystal structure of MtPFD has
shown that PFD resembles the shape of a jellyfish
with six tentacles. This result has shown that the
distal regions of the coiled coils that expose hydrophobic patches are required for multivalent binding
of nonnative proteins.11 This proposed substratebinding mechanism was supported by the microscopic observation that the substrate interaction sites
of PhPFD are at the tips of the coiled coils of each
subunit.6 In addition, we have reported that both the
N- and C-terminal regions of the PhPFD β subunit
are important for its molecular chaperone activity.7
However, the substrate-binding residues that form
the hydrophobic patches and molecular dynamics
(MD) of each subunit during substrate binding are
still unclear. In this study, we determined the crystal
structure of PhPFD at a resolution of 3.0 Å, examined
the substrate-binding mode by MD simulation, and
characterized the substrate binding site of PhPFD.
Results
Determination of the crystal structure of PhPFD
PhPFD is a hexameric complex composed of two
α subunits (151 amino acids) and four β subunits (117 amino acids). The crystal of PhPFD,
with a space group of P21212, contains an αβ2
trimer in the asymmetrical unit. The two β subunits in the asymmetrical unit were designated β1
and β2.
We could not determine the structure of the Nand C-terminal residues of each subunit (residues
1–3 of the α subunit, residues 1–4 and 111–117 of
the β1 subunit, and residues 1–9 and 111–117 of the
β2 subunit) because of poor electron densities
probably due to the disordered structure of the
regions. In addition, several ambiguous electron
densities were observed at residues 4–12, 16, 19, 22,
24, 26, 47, 86, 136–143, 145, and 148 in the α
subunit; residues 7, 9, 12, 14, 16, 19, 27, 98, 99, 101,
103–105, and 108 in the β1 subunit; and residues
12, 14, 17, 30, 95, 98, 101, 105, 106, and 108 in the
β2 subunit. All atoms of the proteins, except for
these residues, were well fitted to the 2Fo − Fc
electron density map. Analysis of the stereochemical quality of the models was accomplished using
the program PROCHECK.12,13 Finally, the structure
was refined to an R-factor of 0.240. Due to the
relatively low resolution of 3.0 Å and disordered
structure, it was difficult to determine the side
chain conformations of the amino acid residues
with confidence. However, this resolution range is
sufficient to discuss the overall main chain conformation and the orientations of the side chains of
the amino acid residues.
1133
Structure and MD Simulation of Archaeal Prefoldin
Overall structure of PhPFD
PhPFD is a hexameric complex of two α subunits
and four β subunits. The structure of PhPFD is a
double-β-barrel assembly, with six long coiled coils
protruding from it like a jellyfish with six tentacles,
and has a large central cavity formed by the six coiled
coils (Fig. 1a). The structure is almost the same as that
of MtPFD. Each coiled-coil tentacle is fully solvated.
Its polar and charged side chains are almost exposed
to the solvent. On the other hand, hydrophobic
residues form a hydrophobic groove between the
two α-helical regions in the same subunit. These
hydrophobic grooves in the cavity are likely to be
responsible for the interaction with the hydrophobic
surface of an unfolded protein. The average B-factor
of the coiled-coil region of each subunit is much
higher than the β-assembly region, and the coiledcoil tentacles have few intersubunit interactions with
other subunits in the hexameric complex, indicating
that each individual coiled coil is highly flexible. This
flexibility must be a favorable for widening the central cavity and capturing various nonnative proteins.
Structures of the α and β subunits
The α and β subunits share a similar architecture,
as shown in Fig. 1b. In the β subunit, the N- and Cterminal regions (residues 5–47 and 67–110 in β1 and
residues 13–47 and 67–110 in β2) form a coiled-coil
structure that is connected by a β-hairpin linker
(residues 55–66 in both β1 and β2) consisting of two
short β strands. The N-terminus of β2 has lost the last
turn of the α helix compared with that of β1. The Nterminal region of β2 could not be built due to ambiguous electron density. The electron density shows
that the helical structural region of the β2 subunit
seems to be deformed by the contact with symmetryrelated molecules in the crystalline state. The Nterminal region of the β2 subunit should take an
α-helical conformation from residue 5, like the β1
subunit, in solution. The orientations of the coiled
coils relative to the β-hairpin regions of β1 and β2
differ from each other. The α subunit also has a
similar basic architecture, containing an α-helical region (residues 4–53 and 97–146), a β hairpin (residues
59–77), and an extra β hairpin (residues 81–95). The
extra β hairpin is involved in the dimerization of the
α subunits in the α2β4 hexamer. The two α helices of
the α subunit are 12 and 14 turns in length. In the
longer C-terminal α helix, the α helix is distorted in
the middle of the helix around residues 120–130,
making a long curved helix that is likely to be
induced to avoid steric hindrance with symmetrical
molecules. The residues located at the distal region
form the hydrogen bond interactions with symmetrical molecules. These observations were also found
in the crystal structure of MtPFD.
Structure of the hydrophobic groove at the distal
region of each subunit
The far distal ends of the coiled-coil tentacles of
both the α and β subunits seem to take partially
random coil structures. Our previous study using
truncated mutants of PhPFD showed that the far
distal ends of coiled-coil tentacles of both the α and
β subunits were not essential for the chaperone
activity.7 Hydrophobic grooves exist in the coiled
coils of both the α and β subunits (Fig. 1c). Since the
hydrophobic grooves are in the central cavity, these
grooves are thought to be involved in binding with a
nonnative substrate. These hydrophobic grooves are
formed by αLeu11, αLeu14, αVal131, and αVal138
in the α subunit and by βVal8, βLeu12, βLeu15,
βLeu103, and βIle107 in the β subunit. In addition,
the poor electron density map showed the Cterminal end of the β subunit, where we could not
add the model to this determined structure, formed
an α-helical structure continuously, suggesting that
βLeu111 also formed part of this hydrophobic
groove in the β subunit. However, the hydrophobic
residues of the α subunit are not orderly facing each
other, compared with those of the β subunit. Thus,
the hydrophobicity of this groove of the β subunit is
higher than that of the α subunit, indicating that the
β subunit is essential for binding nonnative proteins.
This insight is supported by electron microscopy
analyses6 and our previous biochemical studies.7,10
Dynamic structure of PhPFD at 330 K
The crystal structure of PhPFD suggested that
flexibility of the tentacles is important for capturing
and arresting unfolded proteins of various sizes.
Therefore, we performed MD simulation of a PhPFD
molecule at 330 K, which is close to the temperature
for the measurement of chaperone activities. The
initial model of PhPFD was constructed by energy
minimization from the crystal structure obtained in
this study using the program SANDER in the
AMBER suite. MD simulation was performed by
increasing the temperature gradually from 100 to
330 K to avoid destabilizing the simulated structure.
The plots of the r.m.s. fluctuation (i.e., SD) of each
residue in the α subunit and two β subunits show
the atomic fluctuations along the protein structure
during MD simulation at 300 K in the first 200 ps
and at 330 K in the next 1 ns (Fig. 2a). The residues of
the coiled-coil regions on both the α and β subunits
show a high level of flexibility compared with those
of the β-assembly region. In addition, short segments composed of residues 54–59 in the α subunit
and residues 48–53 in the β subunit exhibit a higher
level of flexibility. These regions in the connecting
linkers between the α-helical and β-assembly regions seem to play the role of a “hinge” and induce
flexible motions of the coiled coils of all subunits.
Figure 2b shows the superimposed structure of the
initial model and simulated model at 330 K.
Although the structure composed of a double-βbarrel assembly and six long coiled coils is kept, the
positions of the coiled coils have drastically changed. The coiled-coil regions of the β1′ (β1 in another
asymmetrical unit) and β2 subunits move forward
toward each other, and the distance between them
was reduced by ∼10 Å, resulting in weak van der
1134
Structure and MD Simulation of Archaeal Prefoldin
Fig. 2. Fluctuation of PhPFD during MD simulation. (a) Fluctuation along the whole trajectory of the Cα atoms of
PhPFD during MD simulation at 300 K for 200 ps (solid line) and at 330 K for 1 ns (dashed line). (b) Superimposed
structure of the initial model (orange) and simulated model (blue). The simulated model structure is the result of the 1 ns
of MD calculation at 330 K.
1135
Structure and MD Simulation of Archaeal Prefoldin
Waals contacts. A similar change was also observed
between the β1 and β2′ subunits. Although the α
subunit moves forward toward the β1 subunit, the
distance between the two α subunits was kept at
∼50 Å. These results showed that the coiled-coil
regions of each subunit tend to move rotationally
and that the width of the central cavity was kept
open. On the other hand, r.m.s. fluctuations in the βassembly region of both the α and β subunits were
less than 1.5 Å, showing a low level of flexibility and
a stable structure. Thus, this β-assembly structure
contributes to the maintenance of this unique
structure and the width of the central cavity.
Docking and MD simulation between PhPFD and
a nonnative protein
Archaeal PFDs have a central role in protecting
unfolded proteins from aggregation under various
conditions and transferring them to group II
chaperonins to facilitate refolding. Previous studies
on archaeal PFDs showed that the substrate-binding
region is at the distal region of the coiled coil.
However, the precise substrate-binding residues
and the motion of PFD during the substrate binding
were not well known. To address these problems,
we performed a docking simulation of denatured
insulin into the central cavity of PhPFD.
At first, we prepared the denatured structure of
insulin by MD simulation. The crystal structure of
human insulin [Protein Data Bank (PDB) accession
code 2c8r] was used as the initial structure. MD
simulation was performed in vacuo at 300 K for
1.0 ns, at 350 K for 1.0 ns, and then at 375 K for 7 ns.
In the simulated denatured structure, hydrophobic
residues are exposed at the surface.
The denatured insulin model was placed in the
PhPFD cavity such that the center of the denatured
insulin was at the central position of the PhPFD
cavity, between the Cα atoms of αLeu11 and βLeu111
of β1′, avoiding unusual short contacts with the
PhPFD. After energy minimization for the PhPFD/
nonnative insulin complex using the program
SANDER in the AMBER suite, the complex is used
for the initial model for the docking MD simulation
(Model 1). In addition, we constructed three additional initial models (Models 2, 3, and 4) by changing
the orientation of the denatured insulin in the
PhPFD/insulin complex at 90, 180, and 270 deg
along with the rotation axis, the line between αLeu11
in the α subunit and βLeu111 in the β1′ subunit.
Table 1 shows the numbers of observed interactions in each of the four docking simulations at 1 ns.
The β subunit has many interactions with the
hydrophobic surface of the denatured insulin compared with the α subunit. Almost all of the interactions in the β subunit are in the hydrophobic
groove at the distal region. Since the β subunit
showed a similar binding mode in all MD simulations with different initial orientations of the
denatured insulin, we analyzed the result of the
MD simulation from Model 1 in detail. β subunits
maintain their secondary structure with coiled-coil
tentacles, although their conformation and orientations are changed. In the MD-simulated structure at
1 ns, the hydrophobic groove at the distal region of
the β subunit forms van der Waals contacts with the
denatured insulin, as shown in Fig. 3a and b. The
distances between Cα atoms of the interacting
hydrophobic residues of the β subunits and insulin
are plotted in Fig. 3c. The distances decreased in the
first 500 ps and then reached constant values in the
next 500 ps. This observation indicated that the β
subunit makes a noticeable positional change and
stable hydrophobic interactions with denatured
insulin. On the other hand, with respect to the α
subunit, the results changed by the initial models. At
1 ns of MD simulation from Models 2 and 3, the
middle of the α-helical structure is loosened and the
distal region of the α subunit is detached from
insulin to widen its central cavity to form a favorable
interaction between insulin and the β subunit.
Meanwhile, in the MD simulations from Models 1
and 4, the α subunit maintained its α-helical structure and moved closer to insulin. These observations
indicated that the α subunit might demonstrate
multiple conformational changes to widen its
central cavity for interaction with substrates in
compliance with the location, binding site, and
molecular size of nonnative substrate proteins.
Chaperone activities of PhPFD β-subunit
mutants
The crystal structure and MD simulation showed
that the hydrophobic groove, at the distal region of
the β subunit formed by βVal8, βLeu12, βLeu15,
βLeu103, βIle107, and βLeu111, is the substrate
binding site; three hydrophobic residues located at
the N-terminal helix and the other three located at
the C-terminal helix are facing each other, as shown
in Fig. 1c. The hydrophobic residues of the α subunit
Table 1. The number of observed interactions between PhPFD and the insulin model in the docking MD simulations
Hydrophobic interactions
Hydrogen bonds
Initial model
α
α′
β1′
β2
α
α′
β1′
β2
Model 1
Model 2
Model 3
Model 4
–
3
9
–
2
–
–
–
7 (7)
9 (9)
8 (5)
9 (7)
12 (10)
7 (6)
13 (11)
10 (8)
13
9
15
2
5
–
–
3
5
3
1
4
3
3
6
3
Numbers of interactions involved in the hydrophobic groove of the β subunit, as mentioned in the text, are shown in parentheses.
1136
Structure and MD Simulation of Archaeal Prefoldin
docking simulations. In addition, our previous
study7 showed that the effect of deletion of Nterminal 17 residues of the α subunit was relatively
marginal compared with the deletion of the βsubunit C-terminus. Therefore, we focused on the β
subunit in this study. To elucidate the role of these
residues of the β subunit, we constructed mutants in
which these residues were replaced by Asn (βV8N,
βL12N, βL15N, βL103N, βI107N, and βL111N) and
investigated their chaperone activity by measuring
their ability to prevent thermal aggregation of
insulin and citrate synthase (CS) and spontaneous
refolding of acid-denatured green fluorescent protein (GFP). Analyses by analytical size-exclusion
chromatography and circular dichroism have shown
that the mutants kept their original quaternary
structures (data not shown). Figure 4 and Table 2
show the chaperone activities of wild-type and
mutant PhPFDs. Wild-type PhPFD efficiently protected insulin and CS from thermal aggregation and
prevented the renaturation of GFP. Although the
effects of mutations on protection of insulin from
aggregation were relatively marginal especially in
the N-terminal region, the results showed a similar
tendency with the other two substrates. In the results
for CS and GFP, βL12N and βI107N exhibited a
drastic decrease in chaperone activities. Since these
two residues face each other and are located in the
middle of this hydrophobic groove, the residues,
especially βIle107, are likely to be important for the
chaperone activity and hydrophobicity of this
groove. We then prepared other mutants in which
βIle107 was replaced by other hydrophobic residues,
Ala (βI107A) and Trp (βI107W). βI107A retained
chaperone activities for both CS and GFP. Unexpectedly, βI107W exhibited partially decreased chaperone activity. Insertion of a bulky amino acid at the
central position of the hydrophobic region might
induce structural disorder in the hydrophobic
region. These results showed that residue β107 is
an essential hydrophobic residue for forming a
hydrophobic patch at the distal region of the β
subunit.
Discussion
Substrate binding site of PhPFD
Fig. 3. Details of the interface between the insulin
model and the distal region of PhPFD. (a and b) Resulting
β1′ subunit (a) and β2 subunit (b) after 1 ns of MDsimulated calculation at 330 K. (c) The distance between
the Cα atoms of the interacting hydrophobic residues of
the PhPFD β subunits and the insulin model structure.
also interacted with the denatured insulin. However,
the residues of the α subunit responsible for binding
with the denatured insulin varied among the four
In this study, we determined the crystal structure
of PhPFD and examined the interaction between
PhPFD and nonnative insulin modeled by MD
simulation and mutant analyses. The overall structure of PhPFD is very similar to that of MtPFD. The
asymmetrical structural unit of both PFDs consisted
of one α subunit (151 amino acid residues in PhPFD
and 141 amino acid residues in MtPFD) and two β
subunits (117 amino acid residues in PhPFD and
122 amino acid residues in MtPFD) (Fig. 1).9 However, the length of the α-helical region of the β
subunit of PhPFD differs from that of MtPFD. The
MtPFD crystal was prepared from mutant MtPFD
Structure and MD Simulation of Archaeal Prefoldin
lacking the last seven residues of the β subunit,
which is not critical for its biological activity. On the
other hand, we used a full-length PhPFD for
crystallization and determined its structure. For the
Cα atoms, the structure of nearly all of the full-length
1137
α subunit (4–151 amino acid residues) was determined. In the β subunit, the secondary structure of
the distal end, which is the truncated region in
MtPFD, was disordered because of the symmetrical
packing in the crystalline state. These unfavorable
contacts at the molecular interface are likely to be the
cause of high mosaicity of the crystals. We tried to
obtain the high-resolution structure in vain. In
addition, the region at the distal end of the β2
subunit contacts that of the symmetrically related β1
subunit. The hydrophobic groove is formed at the
distal end, and the grooves of the β2 and symmetryrelated β1 subunits face each other and form
hydrophobic interactions. The hydrophobic interactions seem to cause the collapse of the secondary
structure at the distal end of the β2 subunit. These
interactions may reflect the substrate-binding state
of PhPFD, in which the hydrophobic groove of the
β subunit interacts with the exposed hydrophobic
residues of unfolded proteins. In this observation,
the hydrophobic groove is formed by βVal8, βLeu12,
βLeu15, βLeu103, βIle107, and βLeu111 of the β
subunit. As expected, site-directed mutagenesis analyses showed that these residues of the β subunit are
essentially involved in nonnative substrate binding
and that βIle107 is a critical residue (Table 2). However, βIle107 is not sufficient to capture the unfolded
protein, because a double mutant (βLeu103N/
β111N) showed a drastic decrease in its chaperone
activities (Table 2). Replacement of βIle107 to a bulky
amino acid, tryptophan, should cause structural
disturbance of the region, which results in the
decrease of chaperone activity. Thus, βIle107 might
serve as a central core to form the hydrophobic
region in addition to direct substrate binding. Our
previous study, using a 6-amino-acid (residues 112–
117) truncation mutant from the C-terminus of the β
Fig. 4. Chaperone activity of PhPFD mutants with the
amino acid replacements in the N-terminal helix (a and b)
and C-terminal helix (c and d) of the β subunit. (a and c)
Effects of wild-type and mutant PhPFDs on the aggregation of CS (100 nM at a monomer concentration) were
monitored by measuring the light scattering at 500 nm
with a spectrofluorophotometer at 50 °C with continuous
stirring. CS was incubated in the absence (filled circles) or
presence [in (a), open circles indicate βL8N; open
triangles, βV12N; open squares, βL15N; and filled
squares, wild type; in (c), open circles indicate βI103N;
open triangles, βI107N; open squares, βL111N; and filled
squares, wild type] of PFDs. AU indicates arbitrary units.
(b and d) Effects of wild-type and mutant PhPFDs on the
refolding of GFP at 60 °C were monitored by the
fluorescence at 510 nm with excitation at 396 nm using a
spectrofluorophotometer. Acid-denatured GFP was
diluted in the folding buffer without (spontaneous, filled
circles) or with [in (b), open circles indicate βL8N; open
triangles, βV12N; open squares, βL15N; and filled
squares, wild type; in (d), open circles indicate βI103N;
open triangles, βI107N; open squares, βL111N; and filled
squares, wild type] PhPFDs. The amount recovered is
expressed as the percentage of the fluorescence intensity of
native GFP.
1138
Structure and MD Simulation of Archaeal Prefoldin
Table 2. Relative activities of the mutated PhPFDs for the
protection of insulin and CS from thermal aggregation and
prevention of GFP spontaneous refolding
Substrates
PhPFD
Wild type
βV8N
βL12N
βL15N
βL103N
βI107N
βL111N
βI107A
βI107W
βL103N/L111N
Insulin (%)a
CS (%)a
GFP (%)b
100
68.5
106
104
85.4
38.9
80.3
87.3
90.4
65.6
100
49.7
25.7
85.6
94.4
12.1
78.1
95.5
53.2
5.1
100
26.4
49.4
63.4
65.7
22.9
79.2
72.2
63.1
25.6
a
Relative suppression of insulin and CS aggregation at the
maximum light scattering of insulin and CS against the wild type
was calculated.
b
Relative suppression of fluorescence recovery at 600 s against
the wild type was calculated.
subunit, showed no observable effect on CS aggregation or GFP refolding.7 These results provided
the convincing evidence for a specific interaction
between this hydrophobic groove and unfolded
proteins.
Roles of the α and β subunits of PhPFD
Recently, an electron microscopic study on PhPFD
showed that the substrate-binding mode changes
with the mass of the substrate unfolded protein; the
number of PhPFD coiled coils involved in the
interaction with the unfolded substrates increases
with the size of the denatured protein.6 With an
unfolded protein of a smaller mass as a substrate,
only a pair of PhPFD β subunits was required for
interaction. Furthermore, the unfolded proteins are
not confined inside the cavity formed by PhPFD
tentacles and rather protrude from it. In this study,
we performed MD simulation analyses of a preliminary rough docking of PhPFD with a nonnative
insulin model structure from MD trajectory at 330 K
and carried out four runs with different initial
models. In all cases, we located the nonnative insulin model in the central cavity of PhPFD. These
results indicate that the substrate-binding mode of
PhPFD is different from that observed in the electron
microscopy study, showing that nonnative insulin
was encapsulated inside the central cavity rather
than protruding from it. In the electron microscopy
study, the smallest substrate protein was a lysozyme
with a molecular mass of 14 kDa, while we used
insulin of 5.7 kDa (consisting of a 2.4-kDa α chain
and a 3.3-kDa β chain) for MD simulation. The
difference in binding mode is likely to be due to the
difference in the molecular size of unfolded substrate proteins. The small mass protein can move
into the cavity, which results in a more stable state as
the interaction with PhPFD increases. Therefore, the
effects of mutant activities for insulin were relatively
marginal, compared with CS and GFP (Table 2). We
are also performing docking simulation with denatured GFP (data not shown). The mobility and
binding mode were almost the same as those with
insulin, but the binding GFP was protruded from
the central cavity as observed in the electron microscopy study.4
In addition, both our results and those from the
electron microscopy study showed that the distal
region of the β subunit was always involved in
substrate binding. The mobile β subunit first
interacts with the surface of the unfolded proteins
by the hydrophobic interaction to protect them from
aggregation. Docking MD simulations at 330 K
showed that both α subunits seemed to change the
width of the central cavity depending on the binding
mode. If the α subunit remains in the same conformation during substrate binding, it causes many
unusual contacts between the substrate and PhPFD,
even with smaller mass substrates. Then, two α
subunits of the hexameric complex need to change
their conformations and positions depending on the
size and binding mode of the unfolded protein to
stabilize the PhPFD/substrate complex. Accompanied by the flexible segments of 54–59 in the α
subunit, the α subunit provides the multiplesubstrate-recognition mechanism of PhPFD and
the wide substrate specificities that arrest various
unfolded proteins. Further functional and MD
comparative studies on PhPFD interaction with
other unfolded proteins and mutational analyses in
the α subunit could help us understand the
distribution of function of each subunit.
Materials and Methods
Crystallization and X-ray data collection
The purification of wild-type PhPFD has been already
reported.10 A crystal of PhPFD was grown by the vapor
diffusion method using a protein solution (20 mg/ml) and
a reservoir solution [30% (v/v) polyethylene glycol 400,
100 mM sodium chloride, and 100 mM lithium sulfate in
100 mM 4-morpholineethanesulfonic acid–NaOH, pH 6.5]
at 18 °C. The diffraction data for PhPFD were collected at
100 K using an ADSC CCD detector system on the NW12
beam line in the Photon Factory and the BL44B2 beam line
in the SPring-8 Center. Diffraction data were processed
using the program HKL2000.16 The collected data and
scaling results are listed in Table 3.
Structure determination and refinements
The structure of PhPFD was determined by the
molecular replacement method using the structure of
MtPFD as a probe model with the program Molrep in
the CCP4 program suite.14 Models were corrected on the
2Fo − Fc electron density map using the program Coot,17
and the structure without solvent molecules was constructed using the maximum-resolution data. Calculations
of structural refinements were carried out by the programs
REFMAC518 and PHENIX.19 After several cycles of positional and temperature factor refinements, three sulfate
ions were found and added to the structure. Finally,
1139
Structure and MD Simulation of Archaeal Prefoldin
Table 3. Data collection and refinement statistics
PhPFD
Temperature (K)
Resolution (Å)
No. of measured reflections
No. of unique reflections
Completeness (%)
Rmergeb
Io/σ(Io)
Space group
Cell dimensions (Å)
a
b
c
Structure refinement
Resolution range (Å)
No. of reflections
Rc
Rfreed
r.m.s.d. bond length (Å)
r.m.s.d. bond angle (°)
No. of protein atoms
No. of ligand atoms (SO4)
100
3.0
61,014
10,215
94.1 (77.9)a
5.5 (43.4)a
25.2 (3.6)a
P21212
65.17
98.75
78.62
50–3.0
9396
0.240
0.277
0.002
0.43
2651
15
a
The values for the highest-resolution shell are given in
parentheses (3.11–3.00-Å resolution).
b
Rmerge = ∑hkl∑i|Ii(hkl) − 〈I(hkl)〉|/∑hkl∑Ii(hkl), where Ii(hkl) is the
ith intensity measurement of reflection hkl, including symmetryrelated reflections, and 〈I(hkl)〉 is its average.
c
R = ∑hkl(|Fo|−|Fc|)/Σhkl|Fo|.
d
Rfree was calculated on 5% of the data omitted randomly.
refinement of the structure converged at an R-factor of
0.240 (Rfree = 0.277)20 with good chemical geometry, as
listed in Table 3. Solvent molecules were not included in
the refined structure due to the relatively low-resolution
range of the diffraction data. The refinement statistics are
listed in Table 3.
Preliminary modeling of the PhPFD complex with
nonnative insulin for MD simulation
At first, we prepared the PhPFD model. The possible
side chain positions of the residues of poor electron
density were placed by manual fitting, avoiding unusual
intramolecular short contacts. The coiled-coil structure of
the β2 subunit at the distal region was replaced by an α
helix, which is the same as the β1 subunit. The structures
of both the β1 and β2 subunits at the C-terminal distal
region (residues βLeu111 and βArg112) were added to the
crystal structure as α helix. The N- and C-terminal
residues (residues 1–4 and 113–117 in the β subunit)
were not added to the model structure because these regions are not critical for its biological activity, as previously described.7 After the generation of the PhPFD
model, structure energy minimization calculation was
carried out with the program SANDER in the AMBER
suite22 without X-ray data.
Second, we prepared the random structure of insulin
showing a nonnative-like structure. The crystal structure
of human insulin (PDB accession code 2c8r) was used as
the initial structure. The insulin molecule was subjected to
restrained MD calculations with parm96 of the AMBER 8
force field in vacuo, where the total number of atoms was
756. The van der Waals and electrostatic interactions were
calculated using the MD Server (NEC Corporation, Japan),
a computer designed for MD calculation.23 A cutoff
distance of 1000 Å was applied to evaluate the van der
Waals and electrostatic interactions. The potential energy
minimization of the initial MD system was carried out
using the conjugated gradient method. Under the restrained conditions that only the water molecules were
allowed to move and the proteins were kept frozen, MD
simulation was started at 100 K for 100 ps using the
energy-minimized structure. The system was equilibrated
by gradually increasing the temperature from 100 to 250 K
for 300 ps in total, where all atoms were allowed to move.
After equilibration, the system was heated to 300 K and
then to 350 K for 1.0 ns at each temperature and to 375 K
for 7 ns. After generating the denatured insulin model, we
defined the center position of PhPFD between the Cα
atoms of αLeu11 in the α subunit and βLeu111 in the β1′
subunit, and the line between them was defined as a
rotation axis. We placed the center position of the nonnative insulin at the center position of PhPFD, avoiding
unusual short contacts with PhPFD. The initial model of
the complex was denoted as Model 1, and energy minimization for the PhPFD/nonnative insulin complex was
performed using the program SANDER in the AMBER
suite. This complex was used as the initial model for
docking MD simulation. In addition, we calculated three
other initial models. The orientation of nonnative insulin
in the PhPFD/insulin complex was rotated by 90, 180, and
270 deg along the rotation axis, denoted as Models 2, 3,
and 4, respectively.
MD simulation
The solvent water molecules within 40 Å of the atoms of
the PhPFD/insulin complex were removed to avoid close
atomic contacts. A half-harmonic potential22 was applied
around the spherical water droplet to prevent the water
molecules from evaporating. The ff96 parameter24,25 in the
AMBER force field, along with the TIP3P water model,26
was used in this study. The van der Waals and electrostatic
interactions were calculated using the MD Server (NEC
Corporation).23 A cutoff distance of 1000 Å was applied to
evaluate the van der Waals and electrostatic interactions.
The potential energy minimization of the initial MD
system was carried out using the conjugated gradient
method. Under the restrained conditions that only the
water molecules were allowed to move and the proteins
were kept frozen, MD simulation was started at 100 K for
100 ps using the energy-minimized structure. The system
was equilibrated by gradually increasing the temperature
from 100 to 250 K for 300 ps in total and to 300 K for 200 ps,
where all atoms were allowed to move. After equilibration, the system was heated to 330 K for 1.0 ns. An
additional 1.0-ns run was performed for detailed analyses
at 330 K, which is close to the measurement temperature of
the PhPFD chaperone activities. The trajectories of the
atoms above 300 K were stored every 1 ps. Visual
Molecular Dynamics27 was used for the analyses of structure and molecular motions and for the preparation of
graphical representations.
Preparation of PhPFD mutants
The single and double mutants of PhPFD (βV8N,
βL12N, βL15N, βL103N, βI107N, βI107A, βI107W,
βL111N, and βL103N/L111N) were constructed with a
QuikChange Site-Directed Mutagenesis Kit (Stratagene,
La Jolla, CA) using the plasmid pPhPFD as a template.7
All constructs were verified by DNA sequencing. The
purification of these mutants was the same as for the wildtype PhPFD.7
1140
Insulin from bovine serum and CS from porcine heart
were purchased from Sigma. GFP was purified as previously described.28,29 Thermal aggregation of CS was
monitored by measuring the light scattering at 500 nm
with a fluorophotometer (FP-6500, JASCO) at 50 °C. GFP
refolding was monitored by measuring the fluorescence
of GFP at 510 nm with excitation at 396 nm. The
measurement of PhPFD activities using CS and GFP as
substrates was done as previously described.7,10 Thermal
aggregation of insulin was monitored by measuring the
light scattering at 500 nm with a fluorophotometer (FP6500) at 60 °C with continuous stirring. Monitoring
started after the addition of insulin (15 μM) to 50 mM
Tris–HCl buffer (pH 8.0) containing 20 mM DTT with
1.5 μM wild-type PhPFD or mutant PhPFDs preincubated
at 60 °C.
PDB accession number
The atomic coordinates and structure factors of PhPFD
have been deposited in the PDB21 under accession code
2ZDI.
Acknowledgements
The work reported here is part of the 21st Century
Center of Excellence Program of “Future NanoMaterials” research and education project, which is
financially supported by the Ministry of Education,
Science, Sports, Culture, and Technology through
the Tokyo University of Agriculture and Technology.
This work was also supported by Grants-in-Aid for
Scientific Research on Priority Areas (17028013,
17066005, and 18031010) and a grant from the
National Project on Protein Structural and Functional Analyses from the Ministry of Education,
Science, Sports, and Culture of Japan to K.M. and
M.Y. We thank the beam line scientists of the Photon
Factory and SPring-8 Center for their help with data
collection.
References
1. Gething, M. J. & Sambrook, J. (1992). Protein folding in
the cell. Nature, 355, 33–45.
2. Hansen, W. J., Cowan, N. J. & Welch, W. J. (1999).
Prefoldin–nascent chain complexes in the folding of
cytoskeletal proteins. J. Cell Biol. 145, 265–277.
3. Siegers, K., Waldmann, T., Leroux, M. R., Grein, K.,
Shevchenko, A., Schiebel, E. & Hartl, F. U. (1999).
Compartmentation of protein folding in vivo: sequestration of non-native polypeptide by the chaperonin–
GimC system. EMBO J. 18, 75–84.
4. Vainberg, I. E., Lewis, S. A., Rommelaere, H., Ampe,
C., Vandekerckhove, J., Klein, H. L. & Cowan, N. J.
(1998). Prefoldin, a chaperone that delivers unfolded
proteins to cytosolic chaperonin. Cell, 93, 863–873.
5. Zako, T., Murase, Y., Iizuka, R., Yoshida, T., Kanzaki,
T., Ide, N. et al. (2006). Localization of prefoldin interaction sites in the hyperthermophilic group II chaperonin and correlations between binding rate and
protein transfer rate. J. Mol. Biol. 364, 110–120.
Structure and MD Simulation of Archaeal Prefoldin
6. Martin-Benito, J., Gomez-Reino, J., Stirling, P. C.,
Lundin, V. F., Gomez-Puertas, P., Boskovic, J. et al.
(2007). Divergent substrate-binding mechanisms reveal an evolutionary specialization of eukaryotic
prefoldin compared to its archaeal counterpart.
Structure, 15, 101–110.
7. Okochi, M., Nomura, T., Zako, T., Arakawa, T., Iizuka,
R., Ueda, H. et al. (2004). Kinetics and binding sites for
interaction of the prefoldin with a group II chaperonin: contiguous non-native substrate and chaperonin
binding sites in the archaeal prefoldin. J. Biol. Chem.
279, 31788–31795.
8. Zako, T., Iizuka, R., Okochi, M., Nomura, T., Ueno, T.,
Tadakuma, H. et al. (2005). Facilitated release of substrate protein from prefoldin by chaperonin. FEBS Lett.
579, 3718–3724.
9. Siegert, R., Leroux, M. R., Scheufler, C., Hartl, F. U. &
Moarefi, I. (2000). Structure of the molecular chaperone prefoldin: unique interaction of multiple
coiled coil tentacles with unfolded proteins. Cell, 103,
621–632.
10. Okochi, M., Yoshida, T., Maruyama, T., Kawarabayasi,
Y., Kikuchi, H. & Yohda, M. (2002). Pyrococcus prefoldin stabilizes protein-folding intermediates and
transfers them to chaperonins for correct folding.
Biochem. Biophys. Res. Commun. 291, 769–774.
11. Martin-Benito, J., Boskovic, J., Gomez-Puertas, P.,
Carrascosa, J. L., Simons, C. T., Lewis, S. A. et al.
(2002). Structure of eukaryotic prefoldin and of its
complexes with unfolded actin and the cytosolic
chaperonin CCT. EMBO J. 21, 6377–6386.
12. Laskowski, R. A., Rullmannn, J. A., MacArthur, M. W.,
Kaptein, R. & Thornton, J. M. (1996). AQUA and
PROCHECK-NMR: programs for checking the quality
of protein structures solved by NMR. J. Biomol. NMR,
8, 477–486.
13. Ramachandran, G. N. & Sasisekharan, V. (1968). Conformation of polypeptides and proteins. Adv. Protein
Chem. 23, 283–438.
14. Merritt, E. A. & Murphy, M. E. (1994). Raster3D
Version 2.0. A program for photorealistic molecular
graphics. Acta Crystallogr., Sect. D: Biol. Crystallogr. 50,
869–873.
15. Kraulis, P. J. (1991). MOLSCRIPT: a program to
produce both detailed and schematic plots of protein
structures. J. Appl. Crystallogr. 24, 946–950.
16. Otwinowski, Z. M. & Minor, W. (1997). Macromolecular crystallography, part A. Methods Enzymol. 276,
307–326.
17. Emsley, P. & Cowtan, K. (2004). Coot: model-building
tools for molecular graphics. Acta Crystallogr., Sect. D:
Biol. Crystallogr. 60, 2126–2132.
18. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997).
Refinement of macromolecular structures by the
maximum-likelihood method. Acta Crystallogr., Sect.
D: Biol. Crystallogr. 53, 240–255.
19. Adams, P. D., Grosse-Kunstleve, R. W., Hung, L. W.,
Ioerger, T. R., McCoy, A. J., Moriarty, N. W. et al.
(2002). PHENIX: building new software for automated crystallographic structure determination. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 58, 1948–1954.
20. Brünger, A. T. (1992). Free R value: a novel statistical
quantity for assessing the accuracy of crystal structures. Nature, 355, 472–475.
21. Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G.,
Bhat, T. N., Weissig, H. et al. (2000). The Protein Data
Bank. Nucleic Acids Res. 28, 235–242.
22. Case, D. A., Pearlman, D. A., Caldwell, J. W.,
Cheatham, T. E., III, Ross, W. S., Simmerling, C. L.
1141
Structure and MD Simulation of Archaeal Prefoldin
et al. (2000). AMBER 6. University of California, San
Francisco, San Francisco, CA.
23. Amisaki, T., Toyoda, S., Miyagawa, H. & Kitamura, K.
(2002). Two algorithms designed for realizing efficient
combination of fast multiple method and dedicated
hardware for molecular dynamics simulations. J.
Comput. Chem. Jpn. 1, 73–82.
24. Cornell, W., Cieplak, P., Bayly, C., Gould, I., Merz, K.,
Jr, Ferguson, D. et al. (1995). A 2nd generation forcefield for the simulation of proteins, nucleic-acids, and
organic-molecules. J. Am. Chem. Soc. 117, 5179–5197.
25. Kollman, P., Dixon, R., Cornell, W., Fox, T., Chipot, C. &
Phorille, A. (1997). The development/application of a
‘Minimalist’ organic/biochemical molecular mechanic
force field using a combination of ab initio calculations
and experimental data. In Computer Simulation of
Biomolecular System: Theoretical and Experimental Applications (van Gunsteren, W. F., Weiner, P. K. & Wilkinson,
26.
27.
28.
29.
A. J., eds), Computer Simulation of Biomolecular System:
Theoretical and Experimental Applications, Vol. 3, pp.
83–96, Kluwer Academic Publishers, Dordrecht, The
Netherlands.
Jorgensen, W., Chandrasekhar, J., Madura, J. & Klein,
M. (1983). Comparison of simple potential functions
for simulating liquid water. J. Chem. Phys., 926–935.
Humphrey, W., Dalke, A. & Schulten, K. (1996). VMD:
visual molecular dynamics. J. Mol. Graphics, 14, 33–38.
Iizuka, R., Yoshida, T., Maruyama, T., Shomura, Y.,
Miki, K. & Yohda, M. (2001). Glycine at the 65th
position plays an essential role in ATP-dependent
protein folding by archaeal group II chaperonin.
Biochem. Biophys. Res. Commun. 289, 1118–1124.
Sakikawa, C., Taguchi, H., Makino, Y. & Yoshida, M.
(1999). On the maximum size of proteins to stay and
fold in the cavity of GroEL underneath GroES. J. Biol.
Chem. 274, 21251–21256.