Molecular medicine-Mechanisms of Protein Aggregation

A substantial body of information supports
the idea that protein aggregation
arises from partially folded intermediates
through hydrophobic interactions. The
formation of aggregates has often been
considered as a trivial phenomenon, a
nonspecific association of partially folded
polypeptide chains to form a disordered
precipitate. However, several analyses indicate
that aggregation occurs by specific
intramolecular associations involving the
recognition of a sequence partner in another
molecule rather than in the same
molecule during the folding process. Analyses
of the aggregation mechanisms of
various proteins, such as bovine growth
hormone and phosphoglycerate kinase,
has permitted the identification of specific
sites that are critical in the association.
An elegant demonstration of the specificity
of aggregation was provided by King
Aggregation, Protein 31
and coworkers. The authors showed that
during the in vitro refolding of a mixture
of two proteins, tailspike endorhamnidase
and coat protein from phage P22, no heterogeneous
aggregates were formed. Tailspike
endorhamnidase is a thermostable
trimer whose folding intermediates are
thermolabile and either undergo productive
folding or formmultimeric aggregates
(Fig. 3). The P22 coat protein, which comprises
the capsid shell of phage P22, yields
either a correct fold or ‘‘off-pathway’’ aggregates
upon refolding. Both proteins were
intensively studied by King and coworkers
who first denatured the two proteins
in urea and then chose refolding conditions
such that aggregation competes
with correct folding. Folding and soluble
aggregates of the two proteins were characterized
either separately ormixed together.
No heterogeneous aggregates were found,
clearly indicating that only self-association
of transient refolding molecules occurs in
the formation of soluble multimers.
One mechanism that accounts for the
formation of aggregates during refolding
of multidomain proteins is domain swapping.
This was first suggested by Monod
and later proposed by Goldberg and colleagues
to account for the formation of
aggregates during the refolding of tryptophanase.
The concept was foreshadowed
by the results of Crestfield and coworkers
in 1962. From their experiments based
on chemical modification of bovine pancreatic
ribonuclease, the authors proposed
that the dimer is formed by exchanging
the N-terminal fragments. The term 3D
domain swapping was introduced in 1994
by Bennett and coworkers to describe the
structure of a diphtheria toxin dimer. The
mechanism involves the replacement of
one domain of a monomeric protein by the
same domain of an identical neighboring
Aggregate = inclusion body
Suppressor mutation
action
(I*)
(I)
40 °C
tsf 30 °C
Substitution
Early-folding
intermediate
Native
tailspike
Protrimer
Fig. 3 The folding pathway of the P22 tailspike protein. (From Mitraki, A., King, J. (1992) FEBS Lett.
307, 20–25; reproduced with permission.)
32 Aggregation, Protein
(a)
(b)
(c)
(d)
Fig. 4 Schematic
representation of
domain swapping. (a)
monomeric protein, (b) and (c)
partially unfolded monomers,
(d) domain-swapped dimer.
molecule, thus resulting in an intertwined
dimer or oligomer, as defined by Eisenberg
and colleagues (Fig. 4). When the
exchange is reciprocated, domain-swapped
dimers are formed. However, if the exchange
is not reciprocated but propagated
along multiple polypeptide chains, higher
order assemblies or aggregates may form.
Domain-swapped oligomers are divided
into two types, open and closed. The open
oligomers are linear and have one closed
interface (closed in themonomer) exposed
to the solvent, whereas closed oligomers
are cyclic and do not expose a closed
interface. Eisenberg and coworkers have
defined the structure of the monomer as
the ‘‘closed monomer’’ and the conformation
of the polypeptide chain in the
domain-swapped oligomer as the ‘‘open
monomer.’’
The ability of monomeric proteins to
swap structural elements requires the presence
of a hinge or linker region that
permits the protein to attain the native
fold with parts of two polypeptide chains.
In fact, domain-swapped structures reveal
regions of protein structure that are
flexible. Bergdoll and coworkers have suggested
that a proline in the linker region,
by rigidifying the hinge region in intermediate
states, might facilitate domain
swapping. Baker and colleagues proposed
that strain in a hairpin loop might predispose
a protein to domain swapping.
The possible role of 3D domain swapping
in the evolution of oligomeric proteins
has been discussed in several reviews.
In the past years, the number of known
domain-swapped proteins has increased
and today about 40 such structures are
solved. One common feature of these proteins
is that all the swapped domains
are from either the N-terminus or the
C-terminus of the polypeptide chain. In
this regard, an interesting example arises
from the work of Eisenberg and his group
on the dimerization of ribonuclease A.
This protein forms two types of dimers
upon concentration in mild acid. The minor
dimer is formed by swapping of its
N-terminal α-helix with that of an identical
molecule. The major dimer results
from the swapping of its C-terminal β-
strand. RNase A was also reported to form
trimers. On the basis of the structure of
the N- and C-terminal swapped dimers,
a model was proposed (Fig. 5). This indicates
that two types of swapping can occur
simultaneously in the same oligomer. Further
biochemical studies have supported
this model. A less abundant trimer in
which only the C-terminal β-strand is
swapped and exhibits a cyclic structure
was also found. RNase represents the
Aggregation, Protein 33
first protein found to form both linear
and cyclic domain-swapped oligomers.
This protein also was described to form
tetramers. Models based on the structures
of dimers and trimers were proposed for
these tetramers. Two linear models exhibit
both types of swapping that occur in
one molecule, and a cyclic tetramer shows
the swapping of the C-terminal β-strand
only. A trimeric domain-swapped barnase
was obtained at low pH and high protein
concentration. Crystallographic studies revealed
a structure suggesting a probable
folding intermediate. Domain swapping
was described for the cell cycle regulatory
protein p13suc1, a small protein of 113
amino acids.
Folding studies as well as molecular
dynamics simulations have shown
that domain swapping occurs in the unfolded
state. Eisenberg and his colleagues
have proposed a free energy diagram for
the pathway of domain swapping. The
free energy difference between the closed
monomer and domain-swapped oligomer
is small since they share the same structures
except at the hinge loop, but the
energy barrier can be reduced under certain
conditions making domain swapping
more favorable. Several molecular or environmental
events may favor the formation
of extended domain-swapped polymers.
Genetic mutations introducing a deletion
in the hinge loop can destabilize the
monomeric form of a protein. The replacement
of only one amino acid can also favor
Fig. 5 Domain swapping in
ribonuclease. Ribbon diagram of the
structures of (a) the ribonuclease A
monomer (2.0 A˚ ), (b) the N-terminal
swapped dimer (2.1 A˚ ), (c) the
C-terminal swapped dimer (1.75 A˚ ), (d)
the N- and C-terminal trimer model, and
(e) the cyclic C-terminal swapped trimer
(2.2 A˚ ) (reproduced from Liu et al. Prot.
Sci. 11, 371, 2002 with permission).
(a)
(d) (e)
(b)
(c)
the polymerization of the mutated protein.
Three-dimensional domain-swapped
oligomers are expected to be increasingly
favored as the protein concentration increases.
Thus, a metabolic change that
increases the concentration of a protein
will favor aggregation. Charge effects,
caused either by mutations or by pH
change or salt concentration, can induce
domain swapping; for example, in RNase
A, a decrease in pH, by protonating the
residues involved in hydrogen bonds and
in salt bridges, lowers the energy barrier
of the formation of the open monomer,
hence inducing domain swapping.
There is great diversity of swapped domains,
with different sizes and sequences.
They can consist of entire tertiary domains
or smaller structural elementsmade
of several residues. No specific sequence
motif seems to be involved among the
swapped domains. Three-dimensional domain
swapping has also been proposed as
a mechanism for amyloid formation. This
aspect will be discussed in Sect. 4.2.
34 Aggregation, Protein
As can be seen here, several mechanisms
exist, which lead to the formation
of aggregates. It is recognized that aggregation
results from the association of
incompletely or incorrectly folded intermediates
through hydrophobic interactions.
In the energy landscape of protein folding,
the presence of local minima separated
by an energy barrier allows the accumulation
of intermediates. If the barrier is
high enough, these intermediates cannot
easily reach the native state, and kinetic
competition thus favors the formation
of aggregates.