Discussion: The Brain-Behavior Relationship

REVIEW

Volume transmission and wiring transmission from cellular

to molecular networks: history and perspectives

L. F. Agnati, 1,2

G. Leo, 1 A. Zanardi,

1 S. Genedani,

3 A. Rivera,

4 K. Fuxe

4 and D. Guidolin

5

1 Department of Biomedical Sciences, Section of Physiology, University of Modena, Modena, Italy

2 IRCCS San Camillo, Venezia, Italy

3 Department of Biomedical Sciences, Section of Pharmacology, University of Modena, Modena, Italy

4 Department of Neuroscience, Division of Cellular and Molecular Neurochemistry, Karolinska Institute, Stockholm, Sweden

5 Department of Human Anatomy and Physiology, Section of Anatomy, University of Padova, Padova, Italy

Received 13 October 2005,

accepted 8 December 2005

Correspondence: L.F. Agnati,

Department of Biomedical

Sciences, Section of Physiology,

University of Modena, Via G.

Campi 287, 41100 Modena, Italy.

E-mail: [email protected]

Abstract

The present paper deals with a fundamental issue in neuroscience: the inter-

neuronal communication. The paper gives a brief account of our previous

and more recent theoretical contributions to the subject and also reports new

recent data that support some aspects of our proposal on two major modes

of communication in the central nervous system: the wiring and the volume

transmission. There exist two competing theories on inter-neuronal com-

munication: the neuron doctrine and the theory of the diffuse nerve network,

supported by Cajal and Golgi, respectively (see their respective Nobel Lec-

tures). The present paper gives a brief account of a view on inter-neuronal

communication in the brain, the volume and wiring transmission concept

that to a great extent reconcile these two theories. Thus, the theory of volume

and wiring transmission are summarized and its recent developments that

allow to extend these two modes of communication from the cellular net-

work to the molecular network level is also briefly illustrated. The explan-

atory value of this broadened view is further enhanced by our recent

proposal on the existence of a Global Molecular Network enmeshing the

entire central nervous system. It may be interesting to note that also the

Global Molecular Network theory is reminiscent of the old reticular theory

of Apathy. Finally, the so-called ‘tide hypothesis’ for diffusion of signals in

the brain is briefly discussed and its possible extension to the molecular level

is for the first time introduced. Early indirect evidence supporting volume

transmission in the brain was the discovery of transmitter-receptor mis-

matches. Thus, as an experimental part of the present paper a new approach

to evaluate transmitter-receptor mismatches is given and evidence for inter-

relationships between temperature micro-gradients and mismatches is pro-

vided.

Keywords central nervous system, D1 receptor, extra-cellular matrix,

Global Molecular Network, tide hypothesis, uncoupling proteins, volume

transmission, wiring transmission.

From the dawn of modern neuroscience the inter-

neuronal communication in the brain has been the

subject of hot debates (Shepherd 1991). In fact, Cajal’s

neuron doctrine (i.e. of a synaptic organization of the

neuronal circuits) antagonized Golgi’s idea of a diffuse

nerve network or, in other words, the contiguity

Acta Physiol 2006, 187, 329–344

� 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x 329

between neurons (Cajal) was in opposition with the

continuity between neurons (Golgi) as hypotheses to

explain inter-neuronal communication in the brain.

However, already from the very beginning (Golgi 1891)

Camillo Golgi pointed out that the material contact

between neurons is not necessary for their communica-

tion by electrical signals, statement that he based on

Volta’s studies on the second class of conductors (i.e.

electrolytic solutions such as the body fluids). Later on

(1942) this type of communication was described in

some particular neuro-anatomical structures by Arvan-

itaky (Arvanitaki 1942, Krnjevic 1986), who named it

ephaptic transmission (from the Greek word ‘to touch’

even if in the reality no ‘touch’ occurs). More recently, it

has been demonstrated that ephaptic interactions

between a cell body and axons or dendrites can be, in

principle, more significant than ephaptic interactions

among axons in a fibre tract. Extra-cellular action

potentials outside axons are small in amplitude and

spatially spread out, while they are larger in amplitude

and much more spatially confined near cell bodies (Holt

& Koch 1999).

Thus, Golgi’s assessment has been proved and it has

certainly several theoretical and practical implications.

As a matter of fact, the experimental studies on the

ephaptic transmission together with the data on the

extra-cellular space (ECS) as communication channels,

via ion-currents, between neurons (Nicholson et al.

1977) have been important points that we have consid-

ered in formulating a broader view of inter-neuronal

communication. However, it should be kept in mind,

that a new conceptual frame is always the result of

contributions of a very large number of investigators

that is possible to cite properly only in a paper entirely

dedicated to the history of that scientific theory. The

scope of the present paper is different, and hence we will

confine ourselves to cite only some authors.

Thus, only the work of Guillemin, Descarrier, Schmitt,

Nicholson, Vizi, Sykova and Bach-y-Rita will be men-

tioned. The main points of their proposals, as formulated

before our original proposal (Agnati et al. 1986) have

been presented in an extensive paper in the WGC

Symposium held in Stockholm in 1989 (Fuxe & Agnati

1991). However, some more recent results of these and

other authors, when needed, will be reported in the text.

As starting point it may be useful to briefly comment

the different proposals. Most of these proposals were

concerned with the inter-neuronal communication and

all of them were dealing with the prototype of inter-

neuronal communication, namely the synaptic trans-

mission. As a matter of fact, most of them used in an

open or hidden fashion the dichotomous classification:

synaptic vs. non-synaptic transmission.

It should be noted that beside the logical limits of

using a negative term in a classification (see Jevons

1890), several modes of communications in the central

nervous system (CNS) were not considered, such as gap-

junctions and, above all, specialized close membrane

juxta-positions where not only electrical, but also

chemical signals can be exchanged (Shepherd 1991,

Nagy et al. 2004) and usually only two types of signals

were considered, namely neuro-transmitters and ions

hence, other chemical signals such as trophic factors

(Ojeda et al. 2003) and enzymes (Vergnolle et al. 2003)

were neglected. Hence, other relevant physical signals

such as thermal signals were not considered at all

(Agnati et al. 1994, 2004a, Horvath et al. 1999, Fuxe

et al. 2005).

In the 1980s, and the following years, we proposed a

broader view based on two main modes of communi-

cation: the ‘wiring transmission (WT)’ and the ‘volume

transmission’ (VT), where communication takes place

via the ECS. Furthermore, it was clearly pointed out

that these two modes of communication allow not

simply the ‘interneuronal communication’ but rather

the ‘intercellular communication’ in the CNS. Later on

(Agnati & Fuxe 2000), we underlined the importance of

this aspect, and introduced the concept of ‘complex

cellular networks’ (CCN). As discussed below, in the

CCN, cells can exchange information thanks a wide

variety of chemical and physical signals. On this basis,

further developments of our theoretical frame have

recently been given with extension of the VT and WT

concept to the molecular networks (Agnati et al. 2002,

2004b, 2005a).

Theoretical aspects on the communication

modes in the CNS

Basic definitions

Biological apparatuses are made by cells capable of

producing and recognizing signals. It is, therefore,

possible to distinguish cells that operate as a source of

signals and cells that operate as a target for signals.

Usually, a cell is both source and target of signals and

this feature leads to the creation of a net. Thus,

communication in biological apparatuses occurs in

networks. Two types of networks could be distin-

guished in any apparatus: cellular networks and

molecular networks.

Let’s now focus our attention on cellular and

molecular networks of the CNS:

• Cellular networks are made by any type of cells present in the CNS. Actually, not simply neuronal

networks but rather CCN should be considered.

CCN can be defined as the set of cells of any type

that exchanges signals in a certain volume of brain

tissue and, thanks to this cross-talk, is capable of

330 � 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x

Perspectives of volume transmission Æ L F Agnati et al. Acta Physiol 2006, 187, 329–344

integrating inputs to give appropriate outputs

(Agnati & Fuxe 2000).

• Molecular networks are mainly made up by pro- teins and carbohydrates, which fill up the ECS and

the interior of the cells and interact with each other

at boundaries of these compartments namely at the

plasma membrane levels to form a ‘Global Molecu-

lar Network’ (GMN). Thus, the GMN pervades

the intra- as well as the extra-cellular environment

of the entire CNS (Apathy 1897, Agnati et al.

2005a). From a functional standpoint, proteins are

likely to be the most important building blocks for

all the molecular networks. For the ones located in

the ECS (extra-cellular part of GMN) carbohy-

drate-associated proteins have also an important

scaffolding role (Rauch 2004).

More recently (Agnati et al. 2002, 2004b, 2005a), the

concept of VT and WT has been extended also to

molecular networks. As for any computational net, also

for CCN and GMN it is possible to distinguish: nodes,

edges and communication signals (see Fig. 1):

• Nodes are elements where elaborations of inputs occur. As far as the CCN are concerned, nodes are

cells or, in a more integrative view of the CNS,

‘local circuits’ (Rakic 1975). As far as the molecu-

lar networks are concerned, nodes are molecules or

molecule aggregates that under the triggering

action of an input signal carry out biochemical

operations.

• Edges are channels where a communication pro- cess between two nodes occurs. As far as the CCN

are concerned, channels are either parts of cells

(e.g. axons or dendrites) and specialized inter-

cellular contacts (e.g. synapses and gap-junctions),

or diffusion pathways in the ECS. As far as the

molecular networks are concerned, channels are

either molecules, that via inter-molecular physical

contacts allow the transmission of information

between nodes, or diffusion pathways in the

medium interposed between nodes.

• A communication signal is a coded information transmitted along a channel.

The characteristics of the channel allow to distinguish

the VT from the WT:

• VT is characterized by a channel with a poorly defined physical substrate and signal transmission

takes place for diffusion (or vector migration) in

the medium interposed between nodes. ‘Transmis-

sion privacy’, when present, is usually given by the

chemical code (e.g. by the specificity of the neuro-

transmitter/receptor interactions).

• WT is characterized by the transmission of the signal along a channel with a well defined

physical substrate, thus a ‘wire’ links the

source-node with the target-node. ‘Transmission

privacy’, usually present, is given by the fact that

the channel is a private channel. Classically, in

the case of CCN the WT-channel is formed by

an axon and a synapse, in the case of GMN it is

usually formed by proteins (but sometimes also

by carbohydrates and lipids) interconnecting the

two nodes. Thus, the signal transmission appears

Figure 1 Schematic representation of a network. In any network nodes, edges (communication channels), and signals can be

detected. In any apparatus cellular networks and molecular networks could be distinguished. Hence, it has been suggested that a

global molecular network enmeshing the entire CNS is present. Cellular and molecular networks are intermingled and operate in

series and in parallel to allow the integrated function of the CNS. Thus, at the brain level not neuronal networks but rather complex

cellular networks and a molecular network enmeshing the entire CNS should be considered. In cellular networks: nodes are cells

or, in more integrative view of the CNS, ‘local circuits’. Channels are either parts of the cells (e.g. axons or dendrites for

neurons) and specialized inter-cellular contacts (e.g. synapses and gap-junctions), or diffusion pathways in the ECS. In molecular

networks: nodes are molecules or molecule aggregates that under the triggering action of an input signal carry out biochemical

operations. Channels are either molecules, that via inter-molecular physical contacts allow the transmission of information between

nodes, or diffusion pathways in the medium interposed between nodes. For further details see text.

� 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x 331

Acta Physiol 2006, 187, 329–344 L F Agnati et al. Æ Perspectives of volume transmission

as a transient change in the conformation of the

down stream molecules.

Summing up, according to the present hypothesis the

CNS is a computational system with at least two

computing nets, the CCN and the GMN, working

sometimes as parallel circuits, sometimes as serial

circuits and overlapping with each other at cell level

(see Fig. 2). In studying the physiology or pathology of

the brain, it is, therefore, important to consider:

(1) The classical net made by cells and cellular

processes that is the CCN, where a special func-

tional role is given to the neuronal circuits since

they are devoted to a prompt integration of

different types of information to give appropriate

inputs to effectors (e.g. skeletal muscles and

visceral organs).

(2) A GMN made by the extra-cellular matrix (ECM)

molecular net (extra-cellular part of GMN) and by

the intra-cellular molecular nets. The intra-cellular

part of the GMN communicates with the extra-

cellular part of GMN via specialized ‘windows’

(e.g. the lipid rafts). The extra-cellular part of

GMN is made by elements that can carry out a

scaffolding and a computational role. As a matter

of fact, signals can be produced, elaborated and/or

released to be transmitted by the extra-cellular part

of GMN. Among these signals one could mention

several growth factors and hyaluronan fragments

and hyaluronan itself (Bellail et al. 2004, Theodo-

sis et al. 2004).

To understand the complementary functional roles of

these two computing networks, two aspects of the CNS

should be considered namely the location of its cells and

the interconnections among them, and hence the geom-

etry and the connectivity of the neuronal circuits, in one

word the ‘topology’ of the circuit. Both these aspects are

deeply affected by the extra-cellular part of the GMN

which, in turn, is produced and, at least in part,

continuously remodelled by the CCN (Rauch 2004).

VT and WT characteristics

It is possible to summarize the main features of VT and

WT sometimes in absolute terms, but more often by a

comparison between them. Their main features can be

listed as follows:

Main characteristics of VT:

A structurally poorly defined channel, made by the

medium which pervades the network, is always inter-

posed between the source-node and the target-node.

Figure 2 Schematic representation of cellular networks and molecular networks in the brain. Only the extra-cellular molecular

networks are schematically outlined. Membrane associated molecular networks are specialized ‘windows’ for interactions between

extra-cellular and intra-cellular molecular networks. For further details see text.

332 � 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x

Perspectives of volume transmission Æ L F Agnati et al. Acta Physiol 2006, 187, 329–344

Ancillary characteristics of VT:

(1) one source usually gives signals to very many

targets that in some cases are difficult to clearly

identify in view of the fact that no clearly identi-

fiable channel can be traced;

(2) the transmission delay is definitely longer than that

for WT;

(3) the safety of the signal transmission is much lower

than that for WT;

(4) the channel is not a private (dedicated) channel,

hence usually it fulfils also other tasks. A conse-

quence of this feature is the low space cost for VT;

(5) the energy cost for signal transmission is usually

much lower than for WT (Agnati et al. 2005b).

Actually, many times the energy for the signal

transmission is a by-product of taking an ‘oppor-

tunistic’ advantage of energy gradients used for

other purposes within the apparatus or within the

cell.

Main characteristics of WT:

A structurally well-defined channel connects the

source-node with the target-node when the two nodes

are not in a direct physical contact.

Ancillary Characteristics of WT:

(1) One source gives signals to many targets that

usually can be clearly identified in view of the fact

that identifiable channels can be traced. Only when

we consider individual synaptic contacts derived

from the same axon may it be possible to assess

that WT is a ‘one-to-one’ transmission since each

synapse often has its ‘individuality’ (Bennett 2000).

(2) The delay is definitely shorter than that for VT.

(3) The safety of the signal transmission is usually very

high, certainly much higher than that for VT.

(4) The channel is usually a private (dedicated) chan-

nel. A consequence of this feature, and of the

feature mentioned in point 1, is a high space filling

due to the WT-channels axons is observed (Agnati

et al. 2005b). However, in the case of a direct

physical contact between the source-node and the

target-node there is not a real channel, but rather

only interface specializations allowing transmission

of information between the two nodes.

(5) The energy cost for signal transmission is much

higher than that for VT.

VT and WT features for cellular networks and molecular

networks

Cellular networks. Volume transmission: It is in

operation where any type of cell releases chemical

signals or produces a physical signal that diffuses at a

distance, larger than the synaptic cleft, in the ECS

medium interposed between the source and the target

cell where the signal is decoded (Fig. 3).

The following detailed features are well established:

Type of VT-signals. Different classes of VT-signals

should be mentioned:

(a) chemical signals: neurotransmitters, neuromodula-

tors, growth factors, ions (e.g. Ca2+, K+ and H+),

gases (e.g. NO, CO2 and CO), enzymes [e.g.

proteinases on proteinase-activated receptors (PAR)

(Vergnolle et al. 2003, Wang & Reiser 2003)];

(b) physical signals: electrotonic currents, temperature

gradients.

Modalities of VT-signal release. Extra-synaptic release

of transmitters, synaptic spill-over of transmitters,

constitutive (Noel & Mains 1991, Mc Neilly et al.

2003) or reverse uptake release of chemicals in the

extra-cellular fluid (ECF) (Fillenz 1995, Chen & Reith

2000), neuronal generation of electrotonic currents

(local field potentials, LFP) (Ren et al. in press),

generation of temperature gradients by cell groups

containing uncoupling proteins (UCP) (Horvath et al.

1999, see also below).

Pathways for VT-signal migration.

(a) isotropic diffusion in the ECS of the brain;

(b) preferential pathways in the ECS of the brain; this

anisotropic migration can occur mainly along

myelinated nerve bundles and perivascular spaces;

(c) cerebro-spinal fluid (CSF) that allows a vector-

assisted migration.

Energy gradients for VT-signal migration.

(a) concentration gradients;

(b) gradients of electrical potentials (for charged sig-

nals);

(c) pressure gradients (see below the ‘tide hypothesis’);

(d) temperature macro-gradients [i.e. between a brain

active region and a brain inactive region (see

Yablonskiy et al. 2000)] and micro-gradients via

UCP.

It should be noted that pressure and the temperature

gradients beside enabling the ECF renewal (i.e. the

homeostasis of the brain internal milieu), favour VT-

signalling (enhancing release and migration) with obvi-

ously lower costs in terms of energy and space than the

classical synaptic transmission (the prototype of WT).

Furthermore, it should be considered that, as indicated

above in point 2, the temperature macro-gradients by

themselves can operate as signals by acting on molecules

that are highly sensitive to temperature (see also below).

� 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x 333

Acta Physiol 2006, 187, 329–344 L F Agnati et al. Æ Perspectives of volume transmission

Decoding systems for VT-signals.

(a) Private decoding systems: receptors, enzymes, ion

channels, temperature sensitive receptors (Kiyohara

et al. 1990) or, more in general, processes involved

in inter-cellular communication based on chemical

reactions markedly affected by temperature (i.e.

with a temperature coefficient of a reaction

(Q10) > 4) (Kiyohara et al. 1990).

(b) Non-private decoding systems: membrane polar-

ization, chemical reactions involved in inter-cellu-

lar communication affected by temperature

alterations with a Q10 £ 2 (Yablonskiy et al. 2000).

Wiring transmission. The main instances where WT can

be identified are (Fig. 4): synaptic contacts, gap-junc-

tions (Nagy et al. 2004), specialized close membrane

juxta-positions where any movement of substances,

such as ions, metabolites and transmitters, can affect

juxtaposed processes of other cells (Shepherd 1991,

Miyata et al. 1994). It should be noted that mixed

synapses have been described also in mammals. Mixed

synapses provide the structural and functional compo-

nents for both chemical and electrical transmission

within a single synaptic contact on an individual post-

synaptic element (Nagy et al. 2004). For the structural

and functional characteristics of each of these instances,

we refer to previous reviews articles (Nagy et al. 2004).

The signals are usually the chemical signals mentioned

in the introductory section. As far as the physical

signals, the hypothesis is advanced that also tempera-

ture micro-gradients can operate as WT-signals when

they are produced in a highly localized structure (e.g. a

varicosity) which is in close apposition with another

structure where molecules highly sensitive to tempera-

ture are localized.

Molecular networks. Much less is known about the VT

and WT in molecular networks. Thus, presently it is only

possible to mention the best characterized examples:

Figure 3 Schematic representation of the main characteristics of volume transmission (VT) for the cellular networks of the brain.

The drawing illustrates the sources for the VT-signals and the energy gradients that allow their migrations in the extra-cellular space

(ECS) of the brain. It should be mentioned that the volume fraction (a ¼ volume of ECS/volume of tissue) is equal to c. 0.20 (Nicholson & Sykova 1998). For indications about the main VT-signals, see the text.

334 � 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x

Perspectives of volume transmission Æ L F Agnati et al. Acta Physiol 2006, 187, 329–344

(a) The main instances where VT can be identified at

extra-cellular level are the release of trophic signals

from binding sites of the ECM (Soulet et al. 1994,

Cotman et al. 1999). The main instances where VT

can be identified at intra-cellular level is the release

of second messengers (especially calcium ions) in

the cytoplasm (Braunewell 2005).

(b) The main instances where WT can be identified at

extra-cellular level are the interactions of proteins

and hyaluronan with, e.g. integrins inserted in the

plasma membrane or, more generally, protein–

hyluronan–protein interactions (see Day & Shee-

han 2001). The main instances where WT can be

identified at intra-cellular level are the decoding

system of integrins that is mediated via protein–

protein interactions. Another clear example is the

RAS system that via protein–protein interactions

transmit signals from the plasma membrane to the

nucleus (Agnati et al. 2003).

Structure and function of the extra-cellular part

of the GMN

Hyaluronan is simple and unadorned and such a

perfection is unusual in biology and derives from an

important role that hyluronan fulfils in a perfect way: to

be the backbone of several ECM molecular complexes

and to be highly suited for space invading (Day &

Sheehan 2001). Furthermore, it has been pointed out

that several ECM molecules tend to occupy a lot of

space in view of the mutual repulsion of their numerous

negatively charged residues (Sykova 2004). The hyalur-

onan molecule in solution can be pictured as a highly

organized extended dynamic coil that can undergo

sharp kinks, bends and folds while still maintaining

distinctive hydrogen-bond conformations. The charac-

teristic chain reorganization time would be within a

nanosecond timescale. This dynamic flexibility would

give it excellent space-invading properties and, at the

Figure 4 Schematic representation of wiring transmission (WT) for cellular networks of the brain. The morpho-functional

arrangements are schematically illustrated and the main signals employed in the WT are listed. It should be noted that the present

proposal of WT-signals is a further development of the Schmitt’s proposal (see the classical paper by Schmitt 1984) who proposed

the term ‘neuroactive substances’ to denote substances which, in low concentrations, activate important processes within brain cells,

particularly in neuronal targets. Note that when active GAP junctions are continuous via the connexoin built channels coming

together. For further details see text.

� 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x 335

Acta Physiol 2006, 187, 329–344 L F Agnati et al. Æ Perspectives of volume transmission

same time, promote very rapid recovery times after

mechanical perturbation and distortion (Day & Shee-

han 2001).

Thus, hyluronan is one of the fundamental blocks of

the extra-cellular part of GMN which has been

modelled in a simple way by Yamaguchi (2000) with

the proposal of the hyaluronan–lectican–tenascin R

(HLT) model of the brain ECM.

Changes in the building blocks assembled in the net

or binding of ligands to the building blocks can cause a

loosing or a tightening of the net. As far as the building

blocks (see Table 1) of the extra-cellular part of the

GMN it should be considered that:

• Lecticans have different central domains. There- fore, brevican causes a tighter mesh than neurocan;

neurocan causes a tighter mesh than aggregacan.

• Lecticans have different number of attached chon- droitin sulphate chains. The hydratation of the net,

which mainly depends on the amount of chondro-

itin sulphate chains bound to the lectican involved.

In principle, the shorter the lectican the less

chondroitin sulphate chains are attached; there-

fore, less water molecules are held in the net

volume. Therefore, brevican causes less fluidity of

the ECS than aggregacan which has a larger

hydrodynamic volume in view of the high number

of chondroitin sulphate chains.

It is obvious that the size of the mesh and the

hydration of the net deeply affect VT signal diffusion in

the ECS, since they alter the hindrance of the fluidity

(tortuosity) of the ECS. However, besides the control of

VT communication, the extra-cellular part of GMN

also exerts a powerful control of the wiring of the

neuronal circuit (hence of the WT) not only during

development, but also during the entire adult life. This

action of the extra-cellular part of GMN is related to its

control of CNS structural and functional plasticity. The

altered structural plasticity is based on the remodelling

of the circuit, i.e. on a change in the geometry of the

circuit which means a change in the location of the cells

and in the number and location of their interconnec-

tions. In contrast the functional plasticity is based on

the modulation of the functional characteristics of the

connections and mainly on a change in the efficacy of

synaptic and gap-junction transmissions, which are not

necessarily associated with alterations in the geometry

of the circuit.

Summing up, the extra-cellular part of GMN as well

as the intracellular nets are made by building blocks

that create via protein–protein and protein–carbohy-

drate interactions a high-order supra-molecular struc-

ture. The extra-cellular part of GMN has both a

scaffolding role in enabling the appropriate location of

the CNS components and a functional role in cooper-

ating to the maintenance of the microenvironment

around cells and in the creation of preferential diffusion

pathways. The extra-cellular part of the GMN also

affects the communication processes in the cellular

circuits. Intercellular WT (i.e. synaptic contacts, but

also gap junctions and specialized membrane juxta-

positions) is regulated via the control of the cellular

circuit geometry as well as via the control of the cell-to-

cell connection efficacy; intercellular VT (i.e. signal

diffusion) is regulated via the control of the net tightness

and of the ECS medium fluidity, altering tortuosity in

the diffusion pathways where VT signals released by a

cell migrate in the ECS to reach the target cells.

In turn, the CCN affect the GMN molecular compo-

sition and topology both for its extra-cellular and for its

intra-cellular part (Rauch 2004). Thus, a complex

interplay between the molecular and topological organ-

ization of the GMN and the inter-cellular WT and VT

occurs and these interactions lead to an optimal tuning

of the CNS integrative action.

A brief account of the ‘tide hypothesis’

The tide hypothesis enlightens two interrelated phe-

nomena: the renewal of the ECF (i.e. the homeostasis of

the internal milieu of the brain) and the macro-

migration of the VT-signals in the ECS. This hypothesis

is based on the evidence that when arterial pressure

waves reach the Sub-Arachnoid Space (SAS), which is

filled with the CSF, the dilation of cerebral arteries

induces cyclic changes in the hydrostatic pressure inside

of the SAS. It should be noticed that arteries and veins

travel in SAS, and are held against the pia by strands of

connective tissue, before penetrating the brain. As each

small vessel enters or leaves the brain, it carries with it a

sleeve of perivascular space (Virchow–Robin space).

This space extends inward, filled with connective tissue

and ECF, to the level at which the vessel becomes a

capillary. Thus, the SAS is continuous with the

Virchow–Robin spaces and with the ECS of the brain

Table 1 Chemical characteristics of the main lecticans present

in the brain

Lectican

Number of amino

acids of the

central domain

Number of

attachment sites

for glycosaminoglycans

Aggregacan 1486 c. 200

Versican 0 2740 c. 20

Versican 1 1753 c. 15

Versican 2 987 c. 6

Versican 3 c. 0 0

Neurocan 651 c. 7

Brevican c. 300 c. 3

336 � 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x

Perspectives of volume transmission Æ L F Agnati et al. Acta Physiol 2006, 187, 329–344

parenchyma (Nolte 2002). Greitz described the effects

that the cardiac pump has on the brain mechanics by

stating that the brain moves passively like ‘a piston’ due

the alternating pressure gradients created by the heart

and this movement affects CSF movement inside the

ventricular system of the brain (Greitz 1993). These

pressure waves in the CSF produce a ‘push and pull’

movement (as a ‘tide’) of the fluid filling the Virchow–

Robin spaces and thus, according to our hypothesis, of

the ECF in the pericapillary spaces, especially of the

external layers of the cerebral cortex. The tide move-

ments of the ECF can affect the VT communication in

the brain especially with regard to the clearance of

signals over the brain–blood barrier and the pervious-

ness of the VT pathways (Agnati et al. 2004a,b). A

schematic representation of our hypothesis is illustrated

in Figure 5. It should be noticed that the circulation of

the fluid in the ECS of the brain is at least in part (may

be mainly for the most superficial brain volume, i.e.

the one in contact with the SAS and reached by the

Virchow–Robin spaces) similar to the renewal of the

fluid within a sponge.

These macro-effects that, as just described, affect the

brain as a whole, can be traced down to molecular level

since, as reported above, the dynamic flexibility of

hyaluronan would give it excellent space-invading

properties and, at the same time, promote very rapid

recovery times after mechanical perturbation and dis-

tortion (Day & Sheehan 2001). Thus, hyaluronan

behaves like a ‘molecular spring’ that can contribute

to the viscous elastic behaviour of the brain paren-

chyma. Furthermore, lecticans but also several ECM

molecules act like a sponge by binding a large number

of water molecules that can be squeezed out by the

pressure waves. In conclusion, the piston-like move-

ments of the brain cause the sponge-like behaviour of

the brain parenchyma that favour VT-signal migration

(tide hypothesis). Therefore, the tide hypothesis can be

discussed as an epiphenomenon of the molecular

structure of the GMN, in particular of hyluronan,

under the mechanical action of the arterial pulses when

they reach the skull. As indicated in the insert of

Figure 5 the piston-like movements of the brain cause a

slight but appreciable stretch of the parenchyma which

Figure 5 Schematic representation of the tide hypothesis of the fluid movements in the brain (Agnati et al. 2005b). The arterial

pulse as well as the ‘piston like’ movement of the brain cause cyclic pressure oscillations within the SAS. The cyclic pressure

oscillation within the SAS induce ‘tide’ movements of the fluid in Wirchow–Robin spaces, hence in the pericapillary spaces and,

eventually, in the ECF channels, that on one hand favours the renewal of the internal milieu of the brain and on the other hand

favours VT-signal migrations. The insert in the figure illustrates the possible stretching action of the piston-like movements of the

brain on the brain parenchyma. The stretching of the parenchyma can cause the release of ATP from neurons and glial cells

(Lazarowski et al. 2000). ATP mainly works as a WT-signal, but the ADE produced by the enzymatic degradation of ATP mainly

works as a VT-signal. Note that only the pre-junctional features of ATP transmission and its receptors are illustrated. Abbreviations:

ECS, extra-cellular space; SAS, sub-arachnoid space; NT, neurotransmitter; A1 and A2A, receptors for ADE; P2X and P2Y,

receptors for ATP. For further details see text.

� 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x 337

Acta Physiol 2006, 187, 329–344 L F Agnati et al. Æ Perspectives of volume transmission

may induce the release of ATP from cortical neurons

and glial cells (Ahmed et al. 2000, Lazarowski et al.

2000, Neary et al. 2003).The enzymatic degradation of

ATP leads to the local generation of Adenosine (ADE)

which can both acts locally to modulate via A1 or A2A

receptor-activation the ATP effects (Illes & Ribeiro

2004) or can diffuse as a VT-signal to have a wider

modulatory role. Thus, it may be surmised that the

piston-like movements of the brain especially, where

ATP releasing neurons are present may cause also an

overall modulation of neural nets occurring in syn-

chrony with the heart rhythm.

Indirect experimental evidence of VT: the

transmitter-receptor mismatch phenomenon

Several experimental approaches have been provided

supporting the existence of VT (Agnati et al. 2000).

However, one of the first pieces of indirect neurochem-

ical evidence was the demonstration of the so-called

‘transmitter-receptor mismatches’ (Herkenham et al.

1984, Kuhar et al. 1985, Agnati et al. 1986, Jansson

et al. 2002). There is here the spatial uncoupling

between the release sites of a transmitter and the

receptors capable of its recognition and decoding.

In the present paper, we will present a new morph-

ometric approach to evaluate mismatch and we will

apply this method to data from our previous experi-

mental work. Thus, transmitter–receptor mismatches

will be studied in the nucleus accumbens where clear

Dopamine (DA)-terminal vs. DA D1-receptor mismat-

ches have been observed (Jansson et al. 1999). For

details on the immunocytochemical procedure, we will

refer to the Jansson’s paper. It is interesting to note that

this mismatch can be linked to a possible role of UCP in

VT and WT (see also Rivera et al. in press).

Quantitative evaluation of DA-terminal vs. DA

D1-receptor mismatches in nucleus accumbens

by means of computer-assisted image analysis

The basis of our methodology is the modified median

Hausdorff distance (Dubuisson & Jain 1994), which is a

similarity measure between two arbitrary point sets A

and B. It is defined as follows:

hðA; BÞ ¼ median a2A

min b2B

ka � bk;

where ||a)b|| denotes the euclidean distance between a and b. In other words, h(A,B) is the median of the set of

distances taken from any point in A to its nearest point

in B. After obtaining the binary images of the two

patterns, one of them was considered as the reference

pattern and further processed to calculate its ‘distance

transform’ (Russ 1995). This algorithm provides a map

where each background pixel is labelled with a value

equal to its distance from the nearest pixel belonging to

the pattern. The distance of each point of the other

pattern to its nearest in the reference pattern is

evaluated by simply reading out the value the map

exhibited at that location. A first estimate [h(A,B)] of

the median Hausdorff distance is then calculated as the

median of this set of values. A second estimate [h(B,A)]

is obtained by assigning the role of reference pattern to

the second binary image and repeating the whole

procedure. Hence, the maximum between the two

values is considered as the final estimate of the

parameter.

Computer-assisted image analysis was used to evalu-

ate this parameter for nucleus hccumbens tissue sections

processed to visualize DA-terminals and DA-receptors

by immunofluorescence. A spatial mismatch between

the distribution of D1 receptors and that of tyrosine

hydroxylase (TH) positive terminals was observed. The

median Hausdorff distance was found of 30 � 5 lm SD, a value significantly different from zero (P < 0.05,

one-sided t-test).

On the possible role of uncoupling proteins in VT and WT

In the inner membrane of the mitochondria, the UCP

mediate the passive transport of hydrogen ions from the

intermembrane space to the matrix compartment.

Through this process, UCP dissipate the proton gradient

across the inner membrane diminishing the ATP

synthesis and generating an important amount of heat

(Nicholls & Locke 1984, Bouillaud et al. 1985, Palou

et al. 1998, Richard et al. 2001, Horvath et al. 2003a).

UCP have also been demonstrated to reduce the

production of superoxides and increase the efflux of

calcium from the mitochondria (Negre-Salvayre et al.

1997, Arsenijevic et al. 2000). The proteins UCP2 and

UCP3 are members of this protein family and at least

UCP2 is expressed in the CNS (Fleury et al. 1997,

Richard et al. 1998, Horvath et al. 1999, Diano et al.

2000, Lengacher et al. 2004). The role played by UCP2/

3 in the brain has yet to be fully clarified, but it has been

suggested that UCP2 could contribute to the control of

neurotransmission (Horvath et al. 1999, 2003a) and

exert neuroprotective actions (Bechmann et al. 2002,

Diano et al. 2003, Horvath et al. 2003a,b, Paradis et al.

2003). As demonstrated in a previous paper (Rivera

et al. in press) UCP2/3 may enhance VT in the ascend-

ing dopaminergic and noradrenergic terminal systems

especially in the nucleus Accumbens shell where trans-

mitter/receptor mismatches are found (Jansson et al.

2002).

It has been shown that in the caudatus noradrenaline

(NA) and UCP2/3 are co-stored in the same varicosities

338 � 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x

Perspectives of volume transmission Æ L F Agnati et al. Acta Physiol 2006, 187, 329–344

and it has been hypothesized that the thermal gradients

produced by the UCP molecules could enhance NA

extra-synaptic release and NA migration in the ECS

from large varicosities specialized for VT (Rivera et al.

in press). It has also been demonstrated that in the shell

part of nucleus accumbens UCP2/3 immunoreactivity

co-distributes with TH-positive terminals surrounding

areas of intense D1-immunoreactivity but with very few

TH-positive terminals. Thus, where a marked D1/DA

terminal mismatch is present. These results have been

quantified by this new approach (see Fig. 6), where it is

shown that in the dorso-medial and ventro-medial fields

of the shell region of accumbens a high co-distribution

of TH (i.e. sites of DA release) and UCP molecules are

detected (P < 0.05, one-sided t-test). These results can

be interpreted as shown in Figure 7, where it is

indicated that UCP can favour DA diffusion, hence

VT in the basal ganglia. Actually, it may be surmised

that UCP can have a role both in VT and in WT and it

may even be by itself a WT-signal. Let us examine the

panels of Figure 8; a UCP-positive cell group which is

the source of a chemical VT-signal can create a thermal

gradient favouring the migration of a wave of VT-

signals towards a cell group that is target for that VT-

signal (see Fig. 8 upper panel). A group of UCP-positive

terminals can create a thermal gradient modulating WT

and VT by enhancing the neuronal firing rate and hence

the neurotransmitter release (see Fig. 8 middle panel).

Finally, UCP at the level of a terminal bouton can

produce a thermal signal at the level of special mem-

brane juxta-positions that is detected by biochemical

reactions characterized by a high Q10 value (see, e.g.

Kiyohara et al. 1990), hence this localized thermal

gradient works as a WT-signal (see Fig. 8 lower panel).

It should be noted that in the pre-optic area and anterior

hypothalamus neurons have been described with a non-

inactivating sodium channel that is highly temperature

sensitive in the hyperthermic range with a Q10 that can

even reach the value of 7 (Kiyohara et al. 1990).

General discussion

The present paper illustrates a broaden view of our

early proposal of VT and WT as the two modes of

intercellular communication in the brain. It should be

noted that the focus of the communication process with

which we dealt from the beginning was the inter-cellular

and not the inter-neuronal communication. On the basis

Figure 6 Evaluation of the mismatch phenomenon at nucleus Accumbens level for the TH-positive terminals (site of DA-release)

and UCP2/3 immunoreactive structures (for details on the procedure see Rivera et al. in press). Binary images of the TH-positive

terminals and UCP-positive receptors have been obtained by grey level thresholding. The mean grey level exhibited by the back-

ground plus two standard deviations was the threshold value used to identify the DA-terminals and UCP positive structures. The

resulting spatial mismatch is close to zero (high match) in the dorso-medial and ventro-medial fields, while a significant mismatch

(h ¼ 12.5; P < 0.05, one-sided t-test) was shown in the lateral field.

� 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x 339

Acta Physiol 2006, 187, 329–344 L F Agnati et al. Æ Perspectives of volume transmission

of the present proposal, we think that not only the old

controversy between Golgi’s and Cajal’s views is

conciliated, but also that in addition to synaptic

contiguity and reticular continuity, large discontinuities

between interacting cells are present in the brain.

Furthermore, we have revisited the old Apathy’s

proposal of a diffuse fibrillar network interconnecting

the cellular elements of the brain. This proposal has

now a sounder support from the investigations on the

ECM molecular networks. Thus, we have proposed the

existence of a GMN enmeshing the entire CNS (Agnati

et al. 2005a), pointing out that the ECM molecular

networks have a physical and functional relationships

with intra-cellular molecular networks via specialized

cell membrane molecular networks. Thus, it may be

thought that if it is true that nothing is new under the

sun (Bible: Ecclesiaste) it is also to be considered that

very few proposals are entirely new in Science!

The mismatch phenomenon has been considered by

us as an architecture of VT and thus as indirect evidence

in favour of VT. However, it can also be discussed as a

functional arrangement that allows a particular aspect

of the information handling by the CNS. Let us briefly

introduce this aspect. A theoretical model has been

introduced (Montague et al. 1996) where activity in the

cerebral cortex makes prediction about future rewards

and fluctuations in activity of mesencephalic DA neu-

rons could represent errors in these predictions that are

distributed to cortical and subcortical targets. The

model needs an hidden layer to have a temporal delay

that allows to rank the events according to their

temporal sequence. According to a new formulation of

this model spatial mismatch between terminals and

receptors and VT vs. WT transmission from source

neurons to target neurons could provide a simple and

physiologically consistent way to perform such a

computation without the need for a hidden layer.

Due to the different velocities of WT and VT the

signal at the receptor is correlated with the WT-signal

delivered at the source at an earlier time step. Thus,

the combined operations of VT and WT allow a very

effective morpho-functional organization of a circuitry

involved in a fundamental task in the brain, namely

in rewarding. This possible functional meaning of VT

can be added to the previous proposal of our and

other groups. As a matter of fact, it has been

suggested that VT can have a role in mood tone,

hunger, aggression, sleep, artistic performances (Fuxe

& Agnati 1991, Agnati & Fuxe 2000, Bach-y-Rita

2002). We have also proposed that VT could also

Figure 7 Schematic representation of the possible functional role of UCP2/3 in the Caudatus and nucleus Accumbens (shell part) in

favouring the VT. The scheme suggests the existence of compartments in the basal ganglia where VT signalling occurs and the

efficacy of this mode of intercellular communication is affected by the thermal gradients produced by UCP2/3. For further details see

text.

340 � 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x

Perspectives of volume transmission Æ L F Agnati et al. Acta Physiol 2006, 187, 329–344

play a role in learning and the mechanism allowing

such a learning could be simply a modulation of the

ECS channels between the sources and the targets of

the VT-signals. This proposal was based on Turing’s

B type unorganized machine (Turing 1936) that

Turing demonstrated capable of learning via a control

on the perviousness of the channels connecting the

computing elements of the network. Turing (1952)

proposed that a mechanism similar to the one of his

machine could be active in the first periods of human

neonatal life to allow learning of tasks at cerebral

cortex level. It is possible to surmise that a similar

mechanism can be in operation in some brain areas

also during the adult life representing a very effective

strategy for learning (Agnati et al. 2000).

Conclusions

In conclusion, the main theoretical aspects of our

proposal can be summarized as follows:

(1) In the brain as well as in any apparatus of the

organism two types of networks are strictly inter-

mingled: cellular networks and molecular networks

which interact at special sites of the cell mem-

branes. This has led us to propose the existence of a

GMN enmeshing the entire CNS.

(2) Two modes of communications are in operation in

the CNS and in any apparatus of the organism: the

VT and the WT. These two modes are in operation

at cellular network and molecular network level.

(3) VT and WT can be differentiated on the basis of

the characteristics of their communication channel.

(4) VT and WT have different characteristics and are

employed for different functions. However, in

some instances VT can play a vicarious role in

the presence of a WT deficit (see Bach-Y-Rita

2005).

(5) Diffusion of chemical VT-signals in the brain is

made possible by concentration gradients, and

convection of VT signals is favoured by thermal

Figure 8 Schematic representation of the possible different roles of UCP2/3 in the intercellular communication in the brain. Upper

panel: a UCP2/3 positive cell group which is the source of a chemical VT-signal can create a thermal gradient favouring the

migration of a wave of VT-signals towards a cell group that is target for that VT-signal. Middle panel: a group of UCP-positive

terminals can create a thermal gradient modulating WT and VT by enhancing the neuronal firing rate and hence the neurotrans-

mitter release. Lower panel: UCP at the level of a terminal bouton can produce a thermal signal that, at the level of special

membrane juxta-positions, is detected by biochemical reactions characterized by a high Q10 value, hence this localized thermal

gradient works as a WT-signal.

� 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x 341

Acta Physiol 2006, 187, 329–344 L F Agnati et al. Æ Perspectives of volume transmission

gradients (both produced by localized high neur-

onal firing and by UCP) and by the heart-produced

pressure waves within the skull (tide hypothesis).

Furthermore, the mechanical action of the piston-

like movements of the brain may induce the release

of ATP (Lazarowski et al. 2000, Neary et al. 2003)

which in turn can be degraded to ADE. Hence,

heart rhythm can affect the release of WT (ATP)

and VT (ADE) signals by neurons and astroglial

cells.

(6) The hypothesis is introduced that in some special-

ized membrane juxta-positions where UCP are

highly concentrated and biochemical reactions

with a high Q10 are also present, thermal signals

can by themselves operate as WT-signals. This

view is in agreement with recent investigations

demonstrating that UCP2 in the brain has a

localized action and not a global action on animal

metabolism since in knock-out mice no obesity or

reduced response to cold exposure (Arsenijevic

et al. 2000) could be demonstrated. It is therefore

likely that UCP2 is implicated in a local buffering

action of reactive oxygen species (Paradis et al.

2003) and, according to our hypothesis, it can

produce a highly localized thermal signal that may

represent a WT-signal.

(7) We have also proposed that VT could also play a

role in learning and the mechanism allowing such a

learning could be simply a modulation of the ECS

channels between the sources and the targets of the

VT-signals. This proposal was based on Turing’s B

type unorganized machine that Turing demonstra-

ted capable of learning via a control on the

perviousness of the channels connecting the com-

puting elements of the network (Agnati & Fuxe

2000).

In conclusion, this broad proposal on the communi-

cation processes in the CNS, on the interrelations

between cellular networks and molecular networks

and on the information handling by the CNS not only

conciliate the different views on the communication

processes in the brain, but also give a holistic frame to

this fundamental aspect of brain function. Thus, it is

our opinion that this broad theoretical view can suggest

new investigations and hence it may open up a new

understanding of brain physiology and pathology.

This work was supported by grant from IRCCS Istituto di

Neuroriabilitazione Motoria San Camillo, Venezia, Italy and

by grant from MIUR (PRIN 2004) Roma, Italy.

References

Agnati, L.F. & Fuxe, K. 2000. Volume transmission as a key

feature of information handling in the central nervous system

possible new interpretative value of the Turing’s B-type

machine. Prog Brain Res 125, 3–19.

Agnati, L.F., Fuxe, K., Zoli, M., Zini, I., Toffano, G. & Fer-

raguti, F. 1986. A correlation analysis of the regional dis-

tribution of central enkephalin and beta-endorphin

immunoreactive terminals and of opiate receptors in adult

and old male rats. Evidence for the existence of two main

types of communication in the central nervous system: the

volume transmission and the wiring transmission. Acta

Physiol Scand 128, 201–220.

Agnati, L.F., Cortelli, P., Biagini, G., Bjelke, B. & Fuxe, K.

1994. Different classes of volume transmission signals exist

in the central nervous system and are affected by metabolic

signals, temperature gradients and pressure waves. Neu-

roreport 6, 9–12.

Agnati, L.F., Fuxe, K., Nicholson, C. & Sykovǎ, E. 2000. In:

L.F. Agnati, K. Fuxe, C. Nicholson & E. Sykovǎ (eds) Vol-

ume Transmission Revisited: Progress in Brain Research,

Vol. 125. Elsevier Science BV, Amsterdam.

Agnati, L.F., Santarossa, L., Benfenati, F., Ferri, M., Morpurgo,

A., Apolloni, B. & Fuxe, K. 2002. Molecular basis of learning

and memory: modelling based on receptor mosaics. In: B.

Apolloni & F. Kurfes (eds) From Synapses to Rules, pp. 165–

196. Kluwer Academic/Plenum Publishers, New York.

Agnati, L.F., Ferré, S., Lluis, C., Franco, R. & Fuxe, K. 2003.

Molecular Mechanisms and therapeutical implications of

intramembrane receptor/receptor interactions among hepta-

helical receptors with examples from the striatopallidal

GABA neurons. Pharmacol Rev 55, 509–550.

Agnati, L.F., Vergoni, M.V., Leo, G. et al. 2004a. Energy

gradients for VT-signal migration in the CNS: studies on

melanocortin receptors, mitochondrial uncoupling proteins

and food intake. J Endocrinol Invest 27(6 Suppl.), 23–34.

Agnati, L.F., Santarossa, L., Genedani, S. et al. 2004b. On the

nested hierarchical organization of CNS: basic characteristics

of neuronal molecular networks. In: P. Erdi, A. Esposito, M.

Marinaro & S. Scarpetta (eds) Computational Neuroscience:

Cortical Dynamycs, Lecture Notes in Computer Sciences,

pp. 24–54. Springer, Berlin, Heidelberg, New York.

Agnati, L.F., Tarakanow, A.O., Ferre, S., Fuxe, K. & Guidolin,

D. 2005a. Receptor-receptor interactions, receptor mosaics,

and basic principles of molecular network organization:

possible implications for drug development. J Mol Neurosci

26, 193–208.

Agnati, L.F., Genedani, S., Lenzi, P.L. et al. 2005b. Energy

gradients for the homeostatic control of brain ECF compo-

sition and for VT signal migration: introduction of the tide

hypothesis. J Neural Transm 112, 45–63.

Ahmed, S.M., Rzigalinski, B.A., Willoughby, K.A., Sitterding,

H.A. & Ellis, E.F. 2000. Stretch-induced injury alters mi-

tochondrial membrane potential and cellular ATP in cul-

tured astrocytes and neurons. J Neurochem 74, 1951–1960.

Apathy, S. 1897. Das leitende element des nervensystems und

sein topographischen beziehungen zu de zellen. Mitteilungen

aus der zoologischen Station zu Neapel 12, 495–748.

Arsenijevic, D., Onuma, H., Pecqueur, C. et al. 2000. Dis-

ruption of the uncoupling protein-2 gene in mice reveals a

role in immunity and reactive oxygen species production.

Nat Genet 26, 435–439.

342 � 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x

Perspectives of volume transmission Æ L F Agnati et al. Acta Physiol 2006, 187, 329–344

Arvanitaki, A. 1942. Effects evoked in an axon by the activity

of a contiguous one. J Neurophysiol 5, 89–108.

Bach-y-Rita, P. 2002. Volume transmission and brain plasti-

city. Evol Cogn 8, 115–122.

Bach-Y-Rita, P. 2005. Emerging concepts of brain function.

J Integr Neurosi 4, 183–205.

Bechmann, I., Diano, S., Warden, C.H., Bartfai, T., Nitsch, R.

& Horvath, T.L. 2002. Brain mitochondrial uncoupling

protein 2 (UCP2): a protective stress signal in neuronal in-

jury. Biochem Pharmacol 64, 363–367.

Bellail, A.C., Hunter, S.B., Brat, D.J., Tan, C. & Van Meir,

E.G. 2004. Microregional extracellular matrix heterogeneity

in brain modulates glioma cell invasion. Int J Biochem Cell

Biol 36, 1046–1069.

Bennett, MR. 2000. NANC transmission at a varicosity: the

individuality of single synapses. J Auton Nerv Syst 81, 25–

30.

Bouillaud, F., Ricquier, D., Thibault, J. & Weissenbach, J.

1985. Molecular approach to thermogenesis in brown adi-

pose tissue: cDNA cloning of the mitochondrial uncoupling

protein. Proc Natl Acad Sci U S A 82, 445–448.

Braunewell, K.H. 2005. The darker side of Ca2+ signaling by

neuronal Ca2+-sensor proteins: from Alzheimer’s disease to

cancer. Trends Pharmacol Sci 26, 345–351.

Chen, N. & Reith, M.E. 2000. Structure and function of the

dopamine transporter. Eur J Pharmacol 405, 329–339.

Cotman, S.L., Halfter, W. & Cole, J.Y. 1999. Identification of

extracellular matrix ligands for the heparan sulfate pro-

teoglycan agrin. Exp Cell Res 249, 54–64.

Day, A.J. & Sheehan, J.K. 2001. Hyaluronan: polysaccharide

chaos to protein organisation. Curr Opinion Struct Biol 11,

617–622.

Diano, S., Urbanski, H.F., Horvath, B. et al. 2000. Mitoch-

ondrial uncoupling protein 2 (UCP2) in the nonhuman pri-

mate brain and pituitary. Endocrinology 141, 4226–4238.

Diano, S., Matthews, R.T., Patrylo, P. et al. 2003. Uncoupling

protein 2 prevents neuronal death including that occurring

during seizures: a mechanism for preconditioning. Endo-

crinology 144, 5014–5021.

Dubuisson, M.P. & Jain, K. 1994. Modified Hausdorff dis-

tance for object matching. In: M.P. Dubuisson & K. Jain

(eds) Proceedings of IAPR International Conference on

Pattern Recognition, pp. 566–568. ICPR, Jerusalem.

Fillenz, M. 1995. Physiological release of excitatory amino

acids. Behav Brain Res 71, 51–67.

Fleury, C., Neverova, M., Collins, S. et al. 1997. Uncoupling

protein-2: a novel gene linked to obesity and hyperinsuline-

mia. Nat Genet 15, 269–272.

Fuxe, K. & Agnati, L.F. 1991. Two principal modes of elec-

trochemical communication in the brain: Volume vs. wiring

transmission. In: K. Fuxe & L.F. Agnati (eds) Volume

Transmission in the Brain, Novel Mechanisms for Neural

Transmission, Vol. 1. pp. 1–9. Raven Press, New York.

Fuxe, K., Rivera, A., Jacobsen, K.X. et al. 2005. Dynamics of

volume transmission in the brain. Focus on catecholamine

and opioid peptide communication and the role of uncoup-

ling protein 2. J Neural Transm 112, 65–76.

Golgi, C. 1891. Rendiconti Regio Istituto Lombardo, serie 11,

vol. 24, Fascicolo VIII, IX, Milan.

Greitz, D. 1993. Cerebrospinal fluid circulation and associated

intracranial dynamics: a radiologic investigation using MR

imaging and radionuclide cisternography. Acta Radiol

Suppl. 386, 1–23.

Herkenham, M., Edley, S.M. & Stuart, J. 1984. Cell clusters

in the nucleus accumbens of the rat, and the mosaic re-

lationship of opiate receptors, acetylcholinesterase and

subcortical afferent terminations. Neuroscience 11, 561–

593.

Holt, G.R. & Koch, C. 1999. Electrical interactions via the

extra-cellular potential near cell bodies. J Computational

Neurosci 6, 169–184.

Horvath, T.L., Warden, C.H., Hajos, M., Lombardi, A.,

Goglia, F. & Diano, S. 1999. Brain uncoupling protein 2:

uncoupled neuronal mitochondria predict thermal synapses

in homeostatic centers. J Neurosci 19, 10417–10427.

Horvath, T.L., Diano, S. & Barnstable, C. 2003a. Mitochon-

drial uncoupling protein 2 in the central nervous system:

neuromodulator and neuroprotector. Biochem Pharmacol

65, 1917–1921.

Horvath, T.L., Diano, S., Leranth, C. et al. 2003b. Coenzyme

Q induces nigral mitochondrial uncoupling and prevents

dopamine cell loss in a primate model of Parkinson’s disease.

Endocrinology 144, 2757–2760.

Illes, P. & Ribeiro, A.J. 2004. Molecular physiology of P2

receptors in the central nervous system. Eur J Pharmacol

483, 5–17.

Jansson, A., Goldstein, M., Tinner, B. et al. 1999. On the

distribution patterns of D1, D2, tyrosine hydroxylase and

dopamine transporter immunoreactivities in the ventral

striatum of the rat. Neuroscience 89, 473–489.

Jansson, A., Descarries, L., Cornea-Hérbert, V. et al. 2002.

Transmitter-receptor mismatches in central dopamine,

serotonin, and neuropeptide systems. Further evidence for

volume transmission. In: W. Walz (ed.) The Neuronal

Environment: Brain Homeostasis in Health and Disease, pp.

83–101. Humana Press, Totowa.

Jevons, W.S. 1890. W.S. Jevons (ed.) Elementary lessons in

logic: deductive and inductive. Macmillan and Co. Press,

London and New York.

Kiyohara, T., Hirata, M., Hori, T. & Akaike, N. 1990. Hy-

pothalamic warm-sensitive neurons possess a tetrodotoxin-

sensitive sodium channel with a high Q10. Neurosci Res 8,

48–53.

Krnjevic, K. 1986. Ephaptic interactions: a significant mode of

communications in the brain. NIPS 61, 28–29.

Kuhar, M.J., Unnerstall, J.R. & De Souza, E.B. 1985. Receptor

mapping in neuropharmacology by autoradiography: some

technical problems. NIDA Res Monogr 62, 1–12.

Lazarowski, E.R., Boucher, R.C. & Kendall Harden, T. 2000.

Constitutive release of ATP and evidence for major con-

tribution of ecto-nucleotide pyrophosphatase and nucleoside

diphosphokinase to extracellular nucleotide concentrations.

J Biol Chem 275, 31061–31068.

Lengacher, S., Magistretti, P.J. & Pellerin, L. 2004. Quanti-

tative RT-PCR analysis of uncoupling protein isoforms in

mouse brain cortex: methodological optimization and

comparison of expression with brown adipose tissue and

skeletal muscle. J Cereb Blood Flow Metab 24, 780–788.

� 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x 343

Acta Physiol 2006, 187, 329–344 L F Agnati et al. Æ Perspectives of volume transmission

Mc Neilly, A.S., Crawford, J.L., Taragnat, C., Nicol, L. & Mc

Neilly, J.R. 2003. The differential secretion of FSH and LH:

regulation through genes, feedback and packaging. Reprod

suppl 61, 463–476.

Miyata, S., Nakashima, T. & Kiyoara, T. 1994. Structural

dynamics of neural plasticity in supraoptic nucleus of the rat

hypothalamus during dehydration and rehydration. Brain

Res Bull 34, 169–175.

Montague, P.R., Dayan, P. & Sejnowski, T.J. 1996. A frame-

work for mesencephalic dopamine systems based on pre-

dictive Hebbian learning. J Neurosci 16, 1936–1947.

Nagy, J.I., Dudek, F.E. & John, E.R. 2004. Update on con-

nexoin and gap junction in neurons and glia in te mammilian

nervous sistem. Brain Res Rev 47, 191–215.

Neary, J.T., Kang, Y., Willoughby, K.A. & Ellis, E.F. 2003.

Activation of extracellular signal-regulated kinase by stretch-

induced injury in astrocytes involves extracellular ATP and

P2 purinergic receptors. J Neurosci 23, 2348–2356.

Negre-Salvayre, A., Hirtz, C., Carrera, G. et al. 1997. A role

for uncoupling protein-2 as a regulator of mitochondrial

hydrogen peroxide generation. FASEB J 11, 809–815.

Nicholls, D.G. & Locke, R.M. 1984. Thermogenic mechan-

isms in brown fat. Physiol Rev 64, 1–64.

Nicholson, C. & Sykova, E. 1998. Extracellular space structure

revealed by diffusion analysis. Trends Neurosci 21, 207–

215.

Nicholson, C., Bruggencate, G.T., Steinberg, R. & Stöckle, H.

1977. Calcium modulation in brain extracellular micro-

environment demonstrated with ion-selective micropipette.

Proc Natl Acad Sci U S A 74, 1287–1290.

Noel, G. & Mains, R.E. 1991. The order secretion of bioactive

peptides: oldes or newest first? Mol endocrinol 5, 787–794.

Nolte, J. 2002. Ventricles and cerebrospinal fluid. In: J. Nolte

(ed.) The Human Brain. Mosby press, St Louis, MO.

Ojeda, S.R., Prevot, V., Heger, S., Lomniczi, A., Dziedzic, B. &

Mungenast, A. 2003. Glia-to-neuron signaling and the neu-

roendocrine controlof femalepuberty.AnnMed35, 244–255.

Palou, A., Pico, C., Bonet, M.L. & Oliver, P. 1998. The un-

coupling protein, thermogenin. Int J Biochem Cell Biol 30,

7–11.

Paradis, E., Clavel, S., Bouillaud, F., Ricquier, D. & Richard,

D. 2003. Uncoupling protein 2: a novel player in neuro-

protection. Trends Mol Med 9, 522–525.

Rakic, P. 1975. Local circuit neurons. Neurosci Res Program

Bull 13, 295–416.

Rauch, U. 2004. Extracellular matrix components associated

with remodelling processes brain. Cell Mol Life Sci 61,

2031–2045.

Ren, J., Momose-Sato, Y., Sato, K. & Greer, J.J. in press.

Rhythmic neuronal discharge in the medulla and spinal cord

of fetal rats in the absence of synaptic transmission. J Neuro-

physiol 95, 527–534.

Richard, D., Rivest, R., Huang, Q. et al. 1998. Distribution of

the uncoupling protein 2 mRNA in the mouse brain. J Comp

Neurol 397, 549–560.

Richard, D., Clavel, S., Huang, Q., Sanchis, D. & Ricquier, D.

2001. Uncoupling protein 2 in the brain: distribution and

function. Biochem Soc Trans 29, 812–817.

Rivera, A., Agnati, L.F., Horvath, H.L., Valderrama, H., De la

Calle, A. & Fuxe, K. in press. Uncoupling protein 2/3 im-

munoreactivity and the ascending dopaminergic and nora-

drenergic neuronal systems. Relevance for volume

transmission. Neuroscience 137, 1447–1461.

Russ, J.C. 1995. Processing binary images. In: J.C. Russ (ed.)

The Image Processing Handbook. CRC Press, Boca Raton.

Schmitt, F.O. 1984. Molecular regulators of brain function: a

new view. Neuroscience 13, 991–1001.

Shepherd, G.M. 1991. Foudation of Neuron Doctrine, pp.

1–338. Oxford University Press, New York.

Soulet, L., Chevet, E., Lemaitre, G., Blanquaert, F., Meddahi,

A. & Barritault, D. 1994. FGFs and their receptors, in vitro

and in vivo studies: new FGF receptor in the brain, FGF-1 in

muscle, and the use of functional analogues of low-affinity

heparin-binding growth factor receptors in tissue repair. Mol

Reprod Dev 39, 49–55.

Sykova, E. 2004. Extrasynaptic volume transmission and dif-

fusion parameters of the extracellular space. Neuroscience

129, 861–876.

Theodosis, D.T., Piet, R., Poulain, D.A. & Oliet, S.H. 2004.

Neuronal, glial and syinaptic remodeling in adult hypotha-

lamis: functional consequences and role of cell surface and

extracellular matrix adhesion molecules. Neurochem Int 45,

491–501.

Turing, A. 1936. On computable numbers, with an application

to the Entscheidungsproblem. Proc Lond Math Soc 42, 230–

265.

Turing, A.M. 1952. The chemical basis of morphogenesis.

Philos Trans R Soc Lond B 237, 37–72.

Vergnolle, N., Ferazzini, M., D’Andrea, M.R., Buddenkotte, J.

& Steinhoff, M. 2003. Proteinase-activated receptors: novel

signals for peripheral nerves. Trends Neurosci 26, 496–500.

Wang, H. & Reiser, G. 2003. Thrombin signaling in the brain:

the role of protease-activated receptors. Biol Chem 384,

193–202.

Yablonskiy, D.A., Ackerman, J.J. & Raichle, M.E. 2000.

Coupling between changes in human brain temperature and

oxidative metabolism during prolonged visual stimulation.

Proc Natl Acad Sci USA 97, 9819.

Yamaguchi, Y. 2000. Lecticans: organizers of the brain

extracellular matrix. Cell Mol Life Sci 57, 276–289.

344 � 2006 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2006.01579.x

Perspectives of volume transmission Æ L F Agnati et al. Acta Physiol 2006, 187, 329–344