Abstract
The concept of "protein mosaics" and the
importance of a broader concept of allosteric interactions
for protein folding and protein mosaic assemblage
will be discussed. The concept of "pathological
protein mosaics" will be also introduced as a
fundamental cause of some Conformational Protein Diseases
(CPD).
Thus, the difference between protein aggregates (which
can be non toxic, actually even protective [see data
on Lewy bodies (1)]) and pathological protein mosaics
will be introduced. Furthermore, the relevance of
a prion-like mechanism (the so called "Nosferatu's
effect") will be considered since it may cause
not only misfolding of proteins with physiological
conformations, but also formation of growing toxic
aggregates (pathological protein mosaics). Furthermore,
Alzheimer's disease will be examined as a CPD and
data on the formation of the altered amyloid conformations
will be presented and diffusion of small toxic aggregates
of amyloid peptides as pathogenic Volume Transmission
signals (for a definition of Volume Transmission,
see previous paper this issue) will be discussed.
Finally, the general view will be proposed that PCDs
affect Global Molecular Network function and this
early pathological alteration may only subsequently
affect cellular network function.
Introduction
As discussed in previous papers (see also this issue),
computational networks are physical structures made
up by computing units (vertices or nodes) and channels
(edges) that connect these units (2-4). However, brain
is not simply a network of neurons connected by axons
since also astroglia, microglia, ependymal and capillary
endothelial cells take part in the cellular networks
of the brain (5). Cells themselves should be considered
as networks of molecules connected by biochemical
reactions. Thus, brain is a computational system of
networks of networks (6, 7). As far as molecular networks
are concerned, a special role is played by proteins,
whose function is strictly dependent on their three-dimensional
structure (i.e., their conformation). Until a decade
ago, the main focus in understanding protein dependent
diseases was on aberrant behaviour of enzymes ( e.g.,
ostepetrosis due to a deficit in activity of carbonic
anhydrase II) or on the failure of receptor transduction
(see, e.g., the achondroplasia which is due to alteration
in Fibroblast Growth Factor R 3 activity).
Up to now at least 25 pathological conditions have
been detected which depend on altered three-dimensional
structure of proteins and formation of toxic aggregates.
Thus, a new category of diseases, the Conformational
Protein Diseases (PCDs), has been introduced (8-11).
A peculiar characteristic of the diseases listed in
Fig. 1 is that all of them seem to
have as one common characteristic, namely a high content
of β-structure, which leads to formation of aggregates,
even if the native protein was disordered or rich
in β-helical structure (9).
It is our opinion that many more than 25 diseases
will be characterised which are caused by alterations
in protein conformations.
PCDs may, in fact, become one of the major fields
of investigation of molecular medicine for the next
decades.
This assertion is based not only on the fact that
techniques to clarify the molecular bases of diseases
are getting more advanced, but also on the view, as
indicated in the figure 1. and pointed out by Dobson
(12), of the increasing impact of diseases caused
by civilisation and by longevity.
It has been correctly pointed out that Medicine has
to face new pathologies, which have been defined as
"civilisation diseases" and "post-evolutionary
diseases".
The former ones are caused by collateral effects of
the scientific progress that may also have negative
consequences.
Thus, we have had diseases due to unnatural feeding
of cows (mad cow disease), or to pesticides (Parkinson
disease), or to new therapeutic interventions such
as hemodialysis (peripheral organ amyloidosis). The
latter ones are caused by the unnatural prolongation
of the span of the human life, examples are Alzheimer's
disease, cancers etc. A theoretical and epidemiological
aspect that will also be considered in the present
paper is the astonishing discovery that some proteins
can replicate themselves. Hence, they can serve as
structural templates to transform other protein conformations
into a conformation like their own. This was the Nobel
Prize discovery of Prusiner (13, 14). It has also
been discussed whether such a 'structural inheritance'
played a role in the early steps of biological compartmentalisation,
eventually leading to the formation of the primordial
cells (15). In any case a new view on the aetiological
agents has to be proposed and, furthermore, the border
between life and no-life is fainter than ever (Fig.
2).
Protein folding and the concept of Protein
Mosaics
All of the information needed to specify protein native
conformations is contained within its amino acid sequence
(16). However, the pathway of the folding and the
final conformation is deeply affected by the "energy
landscape" where the folding process takes place.
The energy landscape of a protein describes the protein's
potential energy as a function of conformational coordinates.
The depth of the valleys and the barrier heights of
the rugged traps on the energy landscape modulate
the pathways of protein folding that lead from the
unfolded state, through several transition states,
to a few native low energy states (12). Thus, both
pathways of folding as well the more populated transition
states and the preferred native states are highly
dependent on the energy landscape in which the folding
process occurs.
It is apparent that natural protein sequences are
designed to both favour the proper desired fold (positive
design principle) as well as to minimise the possibility
of the formation of undesired folds (negative design
principle) (17).
Several control mechanisms are in operation to avoid
misfolding of monomers and to prevent formation of
unwanted aggregates (12, 18-20). These quality control
mechanisms are extremely efficient (18). In particular,
a class of proteins, the so-called chaperones (20),
help to obtain the proper protein folding and rescue
the proteins folded in an improper conformation. Chaperones
are found in all types of cells from archea to eukarya
and in various compartments of the eukaryotic cell
(21). A schematic representation of some aspects of
the mechanisms involved in control of folding is given
in Fig. 3.
Molecular signalling pathways to a large extent rely
on protein-protein interactions and, hence, on the
formation of more or less stable protein complexes.
As a matter of fact, proteins possess the so-called
Lego Property (6), i.e., the tendency to form high-order
molecular complexes and this process is affected by
the intrinsic dynamics of monomers since some sub-states
of a monomer are more suitable than others to form,
e.g., homo-oligomers rather than hetero-oligomers.
Thus, all proteins, basically, show a great tendency
to aggregate via protein-protein interactions (6,
22), which may occur via different steric (geometrical)
and chemico-physical modalities. The following different
instances should be considered:
a. Two proteins are complementary either as lock and
key (spatial fitting) or in terms of electrostatic
interactions (positive versus negative charges). They
may also have a hydrophilic character interacting
with a hydrophobic microenvironment or they have a
hydrophobic character interacting with a hydrophilic
microenvironment.
b.Two proteins are not complementary (see point a.)
except for two small complementary domains, which
allow the first weak interaction between the two proteins.
This interaction proceeds if this region of contact
induces a sequential rearrangement of the two proteins.
Thus, if a protein-protein induced-fitting process
occurs as according to a "zipper process".
c. Two proteins are not complementary (see point a.),
but they can be rearranged following the binding of
other ligands to one or both of the partner proteins.
Thus, ligands act as allosteric modulators making
the two proteins complementary.
d. Two proteins are not complementary (see point a.),
but they can be bridged by a third protein that can
bind both.
This broad range of possible modalities for protein-protein
interactions is a source of an innumerable variety
of molecular networks at intracellular, membrane and
extra-cellular level (6, 22-25).
It should be noted that the crowded cytoplasmatic
environment can be the cause of the formation of improper
protein aggregates. As a matter of fact, improper
interactions can be triggered by alterations of pH,
temperature, and water contents.
Let us give some details on this last aspect. The
cytoplasm has a molecular crowding as high as 30-40g/100ml
which is affected by water content of the cell and
hence by extra-cellular fluid osmolarity. It has been
estimated that an increase in macromolecular crowding
from 30 to 33% can cause a rise as high as one order
of magnitude in molecular binding affinities (9).
Thus, every cell type is equipped with mechanisms
designed to maintain cellular volume. This may be
one of the reasons why osmolarity is so strictly controlled
in the organism (see, e.g., 26). It should be mentioned
that in humans an osmolarity increase by only 0.02%
can trigger ADH secretion and in the brain hypo-osmolarity
induces the release by astrocytes of ATP, glutamate,
aspartate and taurine, which buffer the osmolarity
change (27).
The function of a protein complex depends at least
on the following features:
a. Selection of the monomers that are clustered in
the complex
b. Localisation of the monomers in the complex
c. Type of the interactions among monomers, such as
simple binding, allosteric interactions, cooperative
behaviour (28-31).
One analogy to illustrate and summarise all the main
features of a protein complex can be the mosaic, as
conceived in art. A mosaic is a picture or a decorative
design made by setting small coloured cubes (tesserae),
made e.g. of stone or marble, into a surface. The
final design of the mosaic (i.e., the meaning of the
mosaic) depends not only by the building blocks (the
tesserae) but also by their localisation and their
interactions as far as colours and shapes are concerned
(see Fig. 4). Thus, we have introduced
the term "Protein Mosaic" as an extension
of the early proposal of receptor mosaics (22, 24,
32-34). Protein mosaics (PMs) have been defined as
high-molecular-weight complexes capable of "emergent
properties", i.e., of functions that could not
be fully anticipated by analysing the characteristics
of the single elements (tesserae) that form the PM
(6, 7, 22, 34).
In other words, a PM is built up to fulfil a task
and this is achieved through the proper localisation
of its monomers (i.e., topology of its tesserae) which
gives rise to a super molecular structure (the "biochemical
design").
In agreement with the early proposal of receptor mosaics
we suggest that:
uA PM is an ordered protein complex, which usually
operates as a node or an edge in a molecular network.
PMs are characterised by the potentially high chemical
variety of monomers (subunits) which are assembled.
Thus, the same monomers can be part of remarkable
different PMs. They show emergent properties.
uA multimeric protein is an ordered assemblage of
subunits (monomers) which may have signalling role
as in receptor proteins (35), carrier role as in haemoglobin
or structural roles as in collagen. Multimeric proteins
are characterised by the low chemical variety of subunits
which are assembled, thus having similar structures.
Moreover, the same monomers are, usually, part of
remarkably similar multimeric proteins. They show
emergent properties.
uA protein aggregate is an usually disordered protein
complex, which may have as emergent properties either
protective or toxic actions (see below). The precise
meaning of PM can be grasped from Fig. 4,
where not only different PMs can be assembled from
the same building block monomers (tesserae of the
mosaic), but above all these PMs are nodes producing
a different output even in presence of the same input
signal if the topology and/or monomer-monomer interactions
(i.e., the intra-mosaic circulation of the information)
are different. If and only if, an aggregate of proteins
is a node and responds to these requisites in a physiological
manner, fulfilling its functional tasks, can it be
defined as a physiological protein mosaic (22-25).
The Lego Property and hence the PM assemblage is conditioned
not only by the energy landscape and the chemico-physical
characteristics of the protein microenvironment (e.g.,
temperature, pH) but also by allosteric modulators.
In this context it is of fundamental importance to
consider whether a broader application of allosteric
control is in operation both in protein folding and
in monomer assemblage. Let us give the classical definitions
of allosteric interactions (28-31) and then suggest
a broader application of the concept. Allosteric interaction
occurs between two topographically distinct binding
sites on the same protein. A classical example is
that of enzymes whereby binding of regulatory molecules
triggers alterations in the structure of the enzyme
in a way that the binding affinity for the substrate
is either enhanced or diminished. Thus, the concentration
of product is regulated (Fig. 5A).
We have proposed that ligand-triggered allosteric
interactions can occur also between domains within
one and the same protein (Fig. 5B)
and may have an important role in protein folding
and steric conformations. Finally, it should also
be considered that allosteric modulators can affect
interactions between (identical or different) proteins
to favour high-order (homo- or hetero-) oligomers
(Fig. 5C) and this mechanism can
be of importance for the formation of PMs and/or,
more generally, protein complexes.
Pathological Protein Mosaics
A high efficiency of quality control mechanisms is
needed since improper folding and/or clustering of
proteins may lead to formation either of toxic aggregates
or of pathological protein mosaics. Pathological PMs
can be defined as protein complexes that do not fulfil
their functional tasks in a molecular network leading
to abnormal responses of the entire molecular network.
The efficiency of quality control mechanisms declines
with age, and hence toxic aggregates and pathological
PMs become increasingly more frequent in aged people,
as well as in presence of genetic alterations. A typical
secondary structure of several monomers in PCDs is
the b-sheet, which is the basic structure to form
amyloid fibrils as observed in Alzheimer's and Parkinson's
disease (9).
The pathological action of a PM may depend on:
-Loss of physiological functions, which can be due
either to the improper conformation of the monomers
or the improper topology of the protein mosaic or,
finally, to the improper functional interactions among
monomers. The final result is the incapability of
the mosaic to fulfil its task.
-Gain of pathological functions, which can be due
either to the formation of an improper output signal
(a qualitatively or quantitatively altered signal)
or to the formation of abnormal molecular devices
as, for example, the formation of improper membrane
ion channels in Alzheimer's disease (9, 12).
The pathological action of a protein aggregate may
depend on:
-The distortion of cellular structures due to the
spatial encumbrance of the aggregate and the following
possible activation of cellular defence mechanisms
(see e.g., the senile plaques in Alzheimer's disease
and the subsequent microglia activation (36)).
-The formation of low molecular weight (hence diffusible)
aggregates, which are toxic to the cells (37-39).
It has been shown that the so-called ADDLs (Aβ
derived diffusible ligands) affect learning and memory
(40, 41) and subsequently induce neuronal apoptosis
(42, 43).
It should be noticed that PCDs can be caused not only
by primary changes in the proteins involved due to
genetic changes, but also by aberrant behaviour of
chaperones, deficits of proteosome function (the ubiquity-proteosome
system is a protein-quality control system which tags
misfolded or damaged proteins for re-folding, or more
commonly, for degradation), overproduction of Reactive
Oxygen Species (ROS) and interactions of small molecules
with proteins leading to alterations of protein structure
(12).
PCDs can be divided into two groups. In the first
group the main characteristic is the presence of amyloidoses
(hence of protein aggregates), with large quantities
of wrongly folded proteins undergoing aggregation,
which may cause destruction of cells and alterations
of intercellular connections; this is, e.g., one aspect
of Alzheimer's disease. In the second group a small
genetic error leads to a misfolded conformation of
the protein which affects its function. This is the
case observed in several types of cancer where mutant
forms of p53 play an important causal role for the
neoplastic cell proliferation (44). In some instances
it is possible to have both features, as is the case
of mutant forms of p53, which can be present in several
carcinomas as amyloid-like aggregates (44).
From above, it is clear that it is not always possible
to make a clear-cut distinction between pathological
PMs and toxic protein aggregates. However, likely
a deeper insight of at least some PCDs can be achieved
by taking advantage of the concept of PM.
Thus, it is possible to postulate that in some instances
PCDs can also be caused by normally folded proteins
that are misassembled, hence by PMs with abnormal
topology and/or circulation of the information. In
fact, PMs that can either not carry out their function
in nodes or produce an abnormal output. In other words,
there may be a toxic gain of function associated with
aggregation (12) or, as mentioned above, a toxic function
associated with an aberrant topology or circulation
of the information within the PM. This is a completely
new field of investigation on PCDs. As mentioned above,
some metabolites can favour the b-sheet conformation
of monomers and formation of pathological aggregates.
This may be the case of Homocysteine (Hcy) since high
Hcy levels have been reported as an important risk
factor for AD (see 45). This action has been until
now considered due to a Hcy potentiation of Aβ-induced
overproduction of ROS (46). However, we have studied
Hcy actions under a different perspective. Our working
hypothesis aimed to test whether Hcy modulated Amyloid
β (Ab) conformation favouring b-fibril formation.
Actually, we have shown by means of mass spectrometry
(in collaboration with dr. Amina Woods, John Hopkins,
Baltimore) that Hcy binds Aβ and triggers conformational
changes in the peptide favouring the formation of
fibrils both in vitro and in vivo (see Fig.
6, 7).
As mentioned above, an intriguing aspect of some
PCDs has been the discovery that some proteins in
the pathological conformation are endowed with the
extraordinary capability to act as templates to transform
the physiological conformation of another protein
with the same amino acid sequence into a pathological
protein (14, 47). Thus, in some cases, the abnormal
protein can demonstrate an "infectious capability"
as in the case of prions (48) (see Fig. 2),
even if it is likely that still unknown factors are
necessary to generate the infectious form (12, 49).
We suggest calling this action the Nosferatu's effect
from the classical movie of Murnau (produced in the
1922) on vampires (see Fig. 8). A
well documented case of a protein capable of Nosferatu's
effect is the prion protein (PrP) (47, 50, 51). Prion
protein is a cell surface glycoprotein that has a
half-life of about 6 hours, which is found especially
in neuronal cells where it may play a role in copper
metabolism and/or signal transduction (52). It has
been demonstrated that the physiological protein can
convert to the lethal scrapie prion protein isoform
(PrPSc), which is rich in b-structure (53) and forms
aggregates with other PrP molecules and destroys neurons
and glial cells. Conversion can be induced by mutations
(genetic or spontaneous), low pH environments and
exogenous PrPSc.
Thus, the pathological isoform is infectious and is
responsible of several diseases such as bovine spongiform
encephalopathy in cattle, and Creutzfeldt-Jacob, Kuru,
and fatal familial insomnia in humans (52, 54). The
main characteristics of the Nosferatu's effect can
therefore be deduced from the prion diseases (see
Fig. 8). A general scheme can be proposed
where also some crucial questions are posed to which
Medicine has to give prompt answers (see Fig.
9).
Discussion
Evolutionary selection has tended to avoid amino acid
sequences, such as alternating polar and hydrophobic
residues that favour β-sheet structure of the
type seen in amyloid fibrils (12). Furthermore, even
if the aggregation process that results in amyloid
fibrils is nucleated in a similar manner to that of
folding, the residues involved are located in different
regions of the sequence from those that nucleate folding.
Such a "kinetic portioning" means that mutations
have been selected by evolution for their ability
to enhance folding at expenses of aggregation (12).
However, a very important question is the following
one: has the high stability of the amyloid fibrils
(pathological epiphenomenon of a sort of "Super-Lego
Property") in other instances important functional
values? The answer is yes! Data have been obtained
demonstrating that b-sheet structures can be linked
to a variety of non-pathogenetic phenomena ranging
from the transfer of genetic information to synaptic
changes associated with memory (9, 55-57). The potential
relevance of the allosteric interactions within a
protein and between proteins for the formation of
PMs has been illustrated in Fig. 5.
This mechanism can be also in operation for the formation
of pathological PMs and/or protein aggregates. In
agreement with this view, it has been reported that
Ab fibril growth shows features that can be explained
by the fact that the last added Aβ monomer is
stabilized only by the addition of the next monomer,
thus suggesting a sequence-selective facilitation
mechanism of Aβ fibril extension that follows
first-order kinetics (58).
This mechanism can be under subtle controls exerted
by ligands (allosteric regulators) that, by modulating
the conformation of domains, can control the Lego
property. This can, at least partly, explain the Hcy
action illustrated in Fig. 6 and
7, favouring the formation of b-fibrils.
Thus, beside catalysts, chaperones and proteosome
action, also small allosteric molecules control protein
folding and protein aggregation and hence, eventually,
the proper formation of protein mosaics and/or multimeric
proteins. Let us discuss the pathological protein
mosaics as a cause of the failure of the Global Molecular
Network (59; see Agnati and Fuxe also this issue).
Biochemical, structural and ultrastructural data suggest
the existence of a global molecular network enmeshing
the entire CNS (59). Thus, the extra-cellular space
is filled up with a net which is continuous at "special
windows" (the Horizontal Molecular Networks)
with intracellular Vertical Molecular Networks (see
previous paper). On this basis, the existence of three
dimensional molecular networks, mainly made by proteins
and carbohydrates has been proposed, which interact
with each other at the boundaries of compartments
such as plasma membranes to form a "Global Molecular
Network" (GMN) that pervades the intra- as well
as the extra-cellular environment of the entire central
nervous system. The GMN is a potentially plastic structure
regulated through several means. For example, its
extra-cellular part is under the remodelling action
of the matrix metalloproteinases and the cells (especially
glial cells) are continuously turning over this component.
In this context, it is interesting to note that most
of the proteins associated with aggregation diseases
are either intact or fragments of proteins that are
secreted or membrane bound (9). It is, therefore,
suggested that it is possible to distinguish:
-Diseases of the extra-cellular component of the GMN.
The most studied example is the formation of senile
plaques in the Alzheimer's disease. Thus, senile plaques
are parts of the GMN that collapse for some reason
in most cases linked to a conformational alteration
in a secreted protein (Ab) or for some alterations
in hyaluronan (60).
-Diseases of the membrane component of the GMN. A
special case is formation of PrPsc that may occur
at low pH values which exists within caveolae (where
the pH < 5). Furthermore, it has been observed
that lipid rafts, which are special platforms for
horizontal molecular networks (31, 61), are necessary
for PrPsc formation (52).
-Diseases of the intra-cellular component of the GMN
which can be exemplified by the synphilin-1A aggregates
(62).
The general view is proposed that, in a first phase,
conformational protein diseases affect mainly one
of the three components of the GMN and this early
alteration, if uncorrected especially by chaperones,
propagates to the other components of the GMN.
This propagation of the insults can also occur via
the diffusion in the extra-cellular space of the brain
of small toxic aggregates (see above the toxic actions
of ADDLs), and thus via a Volume Transmission mode
(for a definition of Volume Transmission, see 5 and
also this issue).
In a second phase, alterations of the GMN affect the
cellular network, whose functional and trophic impairments
pari passu progress with the alterations in the GMN.
Finally, the overt clinical phase becomes evident
with the degenerative changes, which are usually observed
in all conformational protein diseases (59).
Aknowledgements
The present findings on protein conformations and
protein-protein interactions have been obtained thank
to a grant from MIUR, Rome, Italy (PRIN 2004). Clinical
studies on neurodegenerative diseases are in progress
in collaboration with Istituto di Ricovero e Cura
a Carattere Scientifico (IRCCS) San Camillo; Lido,
VE (prof. Battistin, prof. Tonin and dr. Piron).
Prof.
Agnati L.F. 1 - Prof. Genedani S. 1 - Dott. Leo
G. 1 - Dott. Andreoli N. 1 - Dott. Filaferro M. 1
- Dott. Forni A. 2 - Dott. Guidolin D. 3 - Prof. Mora
F. 4 - Prof. Fuxe K. 5
1Department of Biomedical Sciences,
2Department of Chemistry, University of Modena and
Reggio Emilia, Modena, Italy; 3Department of Human
Anatomy and Physiology, Section of Anatomy, University
of Padova, Padova, Italy;
4Prof. Human Physiology Faculty of Medicine, University
Complutense, Madrid, Spain
5Department of Neurosciences, Karolinska Institutet,
Stockholm, Sweden
luigiagnati@tin.it
