Molecular medicine-Protein Folding, Misfolding, and Aggregation

The New View of Protein Folding
The question of the mechanisms of protein
folding has intrigued scientists for
many decades. As early as the 1930s, attempts
to refold denatured proteins were
published, but significant progress began
to be made when Anfinsen successfully
refolded, denatured, and reduced ribonuclease
into the fully active enzyme. In 1973,
he stated the fundamental principle of protein
folding referred to as the Anfinsen
postulate: ‘‘all the information necessary
26 Aggregation, Protein
to achieve the native conformation of a
protein in a given environment is contained
in its amino acid sequence.’’ The
thermodynamic control of protein folding
was considered to be a corollary of the Anfinsen
postulate, meaning that the native
structure is at a minimum of the Gibbs
free energy. This statement was discussed
by Levinthal in a consideration of the short
time required for the folding process in
vitro as well as in vivo. It was concluded that
a random search of the native conformation
among all possible oneswould require
an astronomic time and is therefore unrealistic.
Thus, it is clear that evolution has
found an effective solution to this combinatorial
problem. This is referred to as
the Levinthal paradox and has dominated
discussions for the last three decades.
In order to understand how the polypeptide
chain could overcome the Levinthal
paradox, different folding models were
proposed and submitted to experimental
tests. Kinetic studies were carried out to
follow the folding pathway. A considerable
number of experiments were performed
to detect and characterize the folding intermediates.
A stepwise sequential and
hierarchical folding process in which several
stretches of structure are formed and
assembled at different levels following a
unique route was supported by a majority
of scientists for many years. According
to this view, misfolded species could be
formed from folding intermediates leading
to the formation of aggregates in a kinetic
competition with the correct folding.
Progressively, with the development of
computers, theoretical studies have approached
the folding problem, using simplified
models to take into account the
computational limitations in simulations
of the folding from the random coil to
the native structure. Different methods
were developed using either lattice models
or molecular dynamics simulations. In
the lattice model, the polypeptide chain
is represented as a string of beads on
a two-dimensional square lattice or on a
three-dimensional cubic lattice. The interactions
between residues (the beads)
provide the energy function for Monte
Carlo simulations. In such simplifiedmodels,
the essential features of proteins, that
is, the heterogeneous character (hydrophobic
or polar) of the interactions and the
existence of long-range interactions, were
included to explore the general characteristics
of the possible folds. Lattice models
were first applied to protein folding by Go
and coworkers while simple exact models
were initiated by Dill and his group, and
have been used by several theoreticians.
From the lattice simulations, insights into
possible folding scenarios have been obtained,
providing a basis for exploring the
general characteristics of folding for real
proteins. The exploration of such models
supplies useful information that can be
submitted to experimental tests.
The so-called ‘‘new view’’ has evolved
during the past 10 years from both experiment
and theory with the use of simplified
models. It is illustrated by the metaphor
of the folding funnel introduced in 1995
by Wolynes and coworkers. The model is
represented in terms of an energy landscape
and describes the thermodynamic
and kinetic behavior of the transformation
of an ensemble of unfolded molecules to
a predominantly native state as illustrated
in Fig. 1. According to this model, there
are several micropathways, each individual
polypeptide chain following its own
route. Toward the bottom of the funnel,
the number of protein conformations decreases
as does the protein entropy. The
steeper the slope, the faster the folding.
As written by Wolynes et al., ‘‘To fold,
a protein navigates with remarkable ease
Aggregation, Protein 27
Native state
Unfolded state
Molten globule
states
Energy
Entropy
Q
Fig. 1 Schematic representation of the folding funnel. Q is the number of
native interactions.
through a complicated energy landscape.’’
Thus, a wide variety of folding behaviors
emerge from the energy landscape,
depending on the energetic parameters
and conditions. The folding rate could be
slowed by ripples in the energy landscape
corresponding to local minima populated
by transiently stable intermediates. In a
rugged energy landscape with kinetic traps
formed by energy barriers, the folding will
be even slower.When local energy barriers
are high enough, protein molecules could
be trapped and possibly aggregate.
The new view has progressively replaced
the classical one of a unique sequential
pathway and is now quite generally accepted.
It is similar to the jigsaw puzzle
model proposed in 1986 by Harrison and
Durbin, suggesting the possibility of multiple
folding routes to reach a unique
solution. Many experimental results are
consistent with this view. There is an
increasing amount of evidence showing
that the extended polypeptide chain folds
through a heterogeneous population of
partially folded intermediates in fluctuating
equilibrium. Several alternative folding
pathways have been observed for different
proteins. From the convergence of
theoretical and experimental studies, a
unified view of the folding process has
progressively emerged, also providing an
explanation for the aggregation processes.