Molecular medicine-The Role of Molecular Chaperones

The discovery of a ubiquitous class of
proteins mediating the correct folding in
cellular environment has led to a reconsideration
of the mechanism of protein
folding in vivo. Historically, the term
molecular chaperone was introduced by
Laskyard and coworkers in 1987 to describe
the function of nucleoplasmin,
which mediates the in vitro assembly
of nucleosomes from separated histones
and DNA. The concept was further extended
by Ellis to define a class of
proteins whose function is to ensure
the correct folding and assembly of
proteins through a transient association
with the nascent polypeptide chain. Studies
on heat-shock proteins have widely
contributed to the development of this
concept.
Today, more than 20 protein families
have been identified as molecular chaperones.
Molecular chaperones comprise
Aggregation, Protein 35
several highly conserved families of related
proteins. They can be divided into two
classes according to their size. Small chaperones
are less than 200 kDa, whereas
large chaperones are more than 800 kDa.
During the past few years, a large amount
of biochemical, biophysical, and low- and
high-resolution structural data have provided
mechanistic insights into the machinery
of protein folding as assisted by
molecular chaperones.
Molecular chaperones are involved in
diverse cellular functions. The constitutive
members of the heat-shock protein family
(Hsp70) can stabilize nascent polypeptide
chains during their elongation in
ribosomes. The large cylindrical chaperonins
GroEL in bacteria, mitochondria,
and chloroplasts and the corresponding
TriC in eukaryotes and archaebacteria
provide a sequestered environment for
productive folding. Several chaperones are
stress-dependent; their expression is induced
under conditions such as high
temperatures, which provoke protein unfolding
and aggregation. The members of
the Hsp90 and Hsp100 families, as well
as small Hsp, play a role in preventing
protein aggregation under stress. Chaperone
interactions are also important for
the translocation of polypeptide chains
into membranes.
Within cells, the nascent polypeptide
chain is synthesized sequentially on the
ribosome by a vectorial process. For
many proteins, the rate of this process
is slower than the rate of folding. Synthesis
times range from 20 s for a 400
residue–polypeptide chain in E. coli at
37 ◦C to 10 times as long for such a
chain in an eukaryotic cell.Many unfolded
proteins refold completely in 20 s under
the same conditions. Thus, there is the
possibility for the elongating polypeptide
either to misfold before completion or to
be degraded by proteolytic enzymes. Chaperones
prevent such unfavorable events by
protecting the nascent chain. Hsp70 and
its prokaryotic homolog DnaK recognize
extended hydrophobic regions of the elongating
polypeptides. These interactions are
not specific.Hsp70 andDnaK interactwith
most unfolded polypeptide chains that expose
hydrophobic residues. They do not
recognize folded proteins. Binding and release
of unfolded proteins from Hsp70
are ATP-dependent and require the presence
of various cochaperones such asDnaJ
and GrpE. The basic mechanism ofHsp70
(DnaK in E. coli) is represented in Fig. 6.
In E. coli, DnaJ binds the nascent unfolded
polypeptide, U; then the complex
binds to the ATP-bound state of DnaK.
ATP is hydrolyzed in the ternary complex
DnaJ
ATP
DnaK
ATP ADP
ATP
ATP
Or towards GroEL
ADP
GrpE
Pi; DnaJ
U
U N
Fig. 6 Schematic representation of the basic mechanism of DnaK (see text).
36 Aggregation, Protein
allowing the release of DnaJ and Pi. In
the following step, GrpE acts as an exchange
factor to regenerate the ATP-bound
state of DnaK. The unfolded polypeptide
chain is released into the bulk solution.
Thus, Hsp70 systems bind and release the
polypeptide in an unfolded conformation.
The unfolded protein has the possibility
either to fold or to be transferred to
the GroEL system, as illustrated in Fig. 6.
Significant insights into this mechanism
were obtained from structural data. The
three-dimensional structures of Hsp70
and DnaJ as well as those of a complex
(a)
(b)
Fig. 7 Crystal structure of
GroEL–GroES–(ADP)7 complex
determined by Sigler et al. (a) view
along the axis and (b) view from the top
of the complex. (Reproduced from the
PDB web site.)
DnaK–polypeptide and a complex of GrpE
with the ATP binding domain of DnaK
are known. DnaK and its homologs are
composed of two domains, a C-terminal
domain that binds ATP and an N-terminal
domain that binds peptides. GrpE is a
tight homodimer associated along twolong
helices. It binds DnaK–ATPase domain
through its proximal monomer. DnaJ activates
the ATP hydrolysis by DnaK. It
was shown that a conformational change
may occur upon ATP binding, opening the
polypeptide binding cleft in the polypeptide
binding domain of DnaK. The closed
state may correspond to the ADP-bound
conformation. The ADP-bound state of
DnaK binds the peptide tightly. Peptide
release requires the dissociation of ADP,
which is mediated by GrpE. DnaK then
rebinds ATP.
The GroEL–GroES system acts by a different
mechanism in which the unfolded
protein is sequestered. The chaperonins
are large cylindrical protein complexes.
The crystal structure of E. coli chaperonin
GroEL was determined in 1994 and that
of the asymmetric GroEL–GroES–(ADP)7
complex in 1997 by Sigler and his group.
GroEL consists of two heptameric rings
of 58-kDa subunits stacked back to back
with a dyad symmetry and forming a
porous cylinder (Fig. 7). Each subunit is
organized in three structural domains. A
large equatorial domain forms the foundation
of the assembly and holds the rings
together. It contains the nucleotide binding
site. A large apical domain forms
the end of the cylinder. The apical domain
contains a number of hydrophobic
Aggregation, Protein 37
residues exposed to the solvent. A small
intermediate domain connects the two
large domains. The intermediate segments
have some flexibility allowing a hingelike
opening of the apical domains, which
occurs upon nucleotide binding. These
movements are large and have been visualized
by three-dimensional reconstruction
from cryoelectron microscopy by Sebil and
her group.
GroES is a heptamer of 10 kDa subunits
forming a flexible dome-shaped structure
with an internal cavity large enough
to accommodate proteins up to 70 kDa.
Each subunit is folded into a single
domain containing β-sheets and flexible
loop regions. The loop regions are critical
for the interactions between GroEL and
GroES. It was deduced from electron
microscopy studies that GroES binding
to GroEL induces large movements in
the apical GroEL domains. This provokes
a significant increase in the volume
of the central cavity in which protein
folding proceeds. NMR coupled with the
study of hydrogen-exchange techniques
has indicated that small proteins are
essentially unfolded in their GroEL-bound
states. Mass spectroscopy has revealed
the presence of fluctuating elements of
secondary structure for several proteins.
In a way, the GroEL–GroES system
recognizes nonnative proteins.
The reaction cycle of the GroEL–GroES
system is represented in Fig. 8. The
nonnative protein binds to the apical
domain of the upper ring of GroEL
through hydrophobic interactions. Then,
the equatorial domain of the same ring
binds ATP, and GroES caps the upper
ring, sequestering the protein inside the
internal chamber in which the protein
folding proceeds. The binding of GroES
induces a conformational change in GroEL
and ATP hydrolysis, which is a cooperative
process that produces a conformational
change in the lower ring, allowing it
to bind a nonnative protein molecule.
This promotes subsequent binding of
ATP and GroES in the lower ring, and
the dissociation of the upper complex,
releasing the protein and ejecting GroES.
If the protein has not reached the native
state, it is subjected to a new cycle.
ADP ADP
ATP
ATP ATP ADP ADP
ATP ATP
ADP
ATP
t ∼ 6 s
Inf N
ADP
A
E
I
GroEL
GroES
(I) (II) (III) (IV)
Fig. 8 The reaction cycle of GroEL–GroES. Inf
is the unfolded protein, N the folded one, A is
the apical domain, (in blue), I the intermediate
domain (in red) and E the equatorial domain (in
magenta). (Reproduced from Wang & Weissman
(1999) Nat. Struct. Biol. 6, 597, with permission.)
(See color plate p. xxii).
38 Aggregation, Protein
The hydrolysis of ATP by GroEL is used
only to induce conformational changes
of the chaperone, which permits the release
of the folded protein. The molecular
chaperones, by their transient association
through hydrophobic interactions with
nascent, stress-destabilized, or translocated
proteins, have a role in preventing
improper folding and subsequent aggregation.
They do not interact with folded
proteins. They do not carry information
capable of directing a protein to assume a
structure different from that dictated by its
amino acid sequence. Therefore, molecular
chaperones assist the folding in the
cells without violation of the Anfinsen postulate.
They increase the yield but not the
rate of folding reactions; in this respect
they do not act as catalysts. Furthermore,
themajority of newly synthesized polypeptide
chains in both bacterial and eukaryotic
cells fold spontaneously without the assistance
of molecular chaperones.
Many proteins from prokaryotic and
eukaryotic organisms are produced with
an amino-terminal propeptide, which is
removed by limited proteolysis during
the activation process. Several of these
propeptides consist of a long polypeptide
chain; for example, there are 174 amino
acids in the propeptide of pro-α-lytic protease,
91 in that of procarboxypeptidase
Y, and 77 in that of prosubtilisin. Several
studies have shown that the propeptide
is required for proper folding of these
proteins. The mature enzymes are not
able to refold correctly. They seem to
have kinetic stability only, whereas the
proenzymes have thermodynamic stability.
Since propeptides perform the function
of mediating protein folding, they have
been classified as intramolecular chaperones.
However, this terminology is not
appropriate since the nascent protein is
the proenzyme, not the enzyme that has
undergone proteolytic cleavage. Thus, it
is not surprising that the proenzyme refolds
spontaneously, whereas the mature
protein does not. Indeed, the information
is contained in the totality of the proenzyme
sequence.
Two other classes of proteins play
the role of helpers during protein folding
in vivo: protein disulfide isomerases
(PDIs) and peptidyl–prolyl cis – trans isomerases.
Protein disulfide isomerase is an
abundant component of the lumen of the
endoplasmic reticulum in secretory cells.
The enzymewas discovered independently
in 1963 by two research groups: in rat
and ox by Anfinsen and coworkers, and
in chicken and pigeon pancreas by Straub
and coworkers. Proteins destined to be secreted
enter the endoplasmic reticulum in
an unfolded state. In this environment,
the folding process is associated with the
formation of disulfide bonds, which is
catalyzed by PDI through thiol–disulfide
interchange. The first PDI cDNA was sequenced
in 1985 by Edman et al. It displays
sequence homologies implying a multidomain
architecture. PDI consists of four
structural domains arranged in the order
a, b, b, a, with the b and a domains
being connected by a linker region. Furthermore,
it possesses an acidic C-terminal
extension. The a and a domains contain
the active site motif – W-C-G-H-C-. They
display significant sequence identity to
thioredoxin, a small cytoplasmic protein
involved in several redox functions, and
they have a similar active site sequence.
Recombinants of the a and b domains
have been obtained and studied by highresolution
NMR. The a domain has the
same overall fold as thioredoxin, an α/β
fold with a central core made up of a
five-stranded β-sheet surrounded by four
helices. As in thioredoxin, the active site
is located at the N-terminus of helix
Aggregation, Protein 39
α2. Preliminary NMR data of the a
domain confirm its structural similarity
to the a domain. The b and b domains
have significant sequence similarity to
each other, but no similarity with the a
domain. Nevertheless, NMR studies of
the b domain have indicated a similar
overall fold. From its sequence, it could
be inferred that b also has the same fold.
Neither b nor b contain the active site.
The folding pathway of disulfide-bound
proteins involves isomerizations between
a number of species containing disulfide
bonds. In vitro experimental studies were
performed using the isolated a and a domains,
and the resultswere comparedwith
those obtained with the holoenzyme. It
was concluded that the activity of long
length PDI is not simply the sum of
the activities of the isolated a and a
domains. Using a series of constructs
including nearly every linear combination
of domains, the contribution of each
domain was investigated. It was determined
that the thiol-disulfide chemistry
requires only the a and a domains, and
that simple isomerization requires one of
these in a linear combination including b,
whereas complex isomerization involving
large conformational changes requires all
the PDI domains except the C-terminal
extension. Thus, it appears that the b domain
is the principal peptide binding site,
but all domains contribute to the binding
of larger polypeptide chains holding
them in a partially unfolded conformation
while the catalytic sites acts synergistically
to perform the thiol-disulfide exchange.
Since PDI has binding properties, it has
been proposed that it acts as a molecular
chaperone. However, as underlined by
Freedman and coworkers, this property
does not represent a chaperone activity
and instead reflects its role as a catalyst to
accelerate the formation of native disulfide
bridges during protein folding.
Several gene products with similarity
to PDI have been identified in higher
eukaryotes. All are probably localized in
the endoplasmic reticulum and have thioldisulfide
exchange activity.
In prokaryotes, the disulfide formation
occurs in the periplasm and is catalyzed by
a protein called DsbA, which exchanges
its Cys30–Cys33 to a pair of thiols in
the target protein, leaving DsbA in its
reduced state. The crystal structure of
oxidized DsbA displays a domain with a
thioredoxin-like fold and another domain,
which caps the thioredoxin-like active site
C30-P31-H32-C33, located at the domain
interface. Reoxidation of DsbA is catalyzed
by a cytoplasmic membrane protein
called DsbB, which contains four cysteine
residues essential for catalysis. DsbB
transfers the electrons from the reduced
DsbA to membrane embedded quinones.
The reduced quinones are then oxidized
enzymatically either aerobically or anaerobically.
Thus, DsbA is found in normal
cells in its oxidized state. E. coli also
has a complex reductive system including
another periplasmic protein DsbC,
which is a homodimer. The molecule
consists of two thioredoxin-like domains
with a CxxC motif, joined via hinged
linker helices to an N-terminal dimerization
domain. The hinge regions allow
movement of the active site, and a broad
hydrophobic cleft between the two domains
may bind the polypeptide chain.
Its function consists of reducing proteins
with incorrect disulfide bonds. DsbC is
maintained in its reduced form by a membrane
protein called DsbD, which contains
six essential cysteine residues. Then, the
electrons are transferred to thioredoxin
and ultimately to NADPH by thioredoxin
reductase.
40 Aggregation, Protein
All these enzymes, which catalyze the
pairing of cysteine residues in disulfidebridged
proteins, have functional domains
pertaining to the thioredoxin superstructure.
Another type of enzyme, peptidyl–prolyl
cis–trans isomerases, facilitates the folding
of some proteins by catalyzing the
cis–trans isomerization of X-Pro peptide
bonds. Two classes of unrelated proteins
demonstrate this activity, those that
bind cyclosporin, which are known as cyclophilins,
and those that bind FK506. The
cellular function of these enzymes is important,
since cyclosporin and FK506 are
potent immunosuppressors that regulate
T-cells activation. Both classes of peptidyl–
prolyl isomerases are ubiquitous,
and abundant in prokaryotes and eukaryotes.
The sequences of several members
of each family are known, and the threedimensional
structures of at least one
member of each family have been elucidated
by X-ray crystallography andmultidimensional
NMR. Their role is to accelerate
the cis-trans isomerization of X-pro peptide
bonds when this process is the ratelimiting
step in protein folding. Although
they do not present structural similarity,
both exhibit a hydrophobic binding cleft
favoring the rotamase activity by excluding
water molecules.