Polymodal regulation of stretch-activated Polymodal regulation of stretch-activated cation channels in leech neurons

UNIVERSITA’ DEGLI STUDI DI PISA

CORSO DI DOTTORATO DI RICERCA

IN NEUROSCIENZE DI BASE E DELLO SVILUPPO

“Polymodal regulation of stretch-activated

cation channels in leech neurons”

Settore scientifico disciplinare: BIO-09

RELATORE:

DOTTORANDA:

Prof. Pellegrino Mario

Dott. Barsanti Cristina

PREMESSA: Le parti in corsivo rappresentano schede di approfondimento relative ad alcuni argomenti introduttivi e possono essere pertanto omesse nel caso di una lettura sintetica. Esse tuttavia risultano utili a lettori non introdotti a questo tema.

Il materiale sperimentale contenuto in questa tesi è stato oggetto delle seguenti pubblicazioni:

Comunicazioni a congressi

• Congresso Nazionale Società Italiana di Neuroscienze – Pisa, 26-28 settembre 2003. “Attivazione rapida di canali cationici meccanosensibili in neuroni di Hirudo medicinalis” – C. Barsanti, M. Pellegrino, M. Pellegrini.

• First International Meeting on Evolution, Development and Neurobiology of the Leech - Villeneuve d’Ascq (France), 29 settembre - 3 ottobre 2004. “Stretch-activated cation channels of leech neurons: characterization and role in neurite outgrowth” - M. Pellegrino, C. Barsanti, D. Ricci, M. Pellegrini.

• 55° Congresso Nazionale Società Italiana di Fisiologia – Pisa, 4-7 ottobre 2004. “Effect of pH on stretch-activated cation channels in leech neurons” - C. Barsanti, M. Pellegrini, M. Pellegrino.

• 7º Convegno FISV (Federazione Italiana Scienze della Vita) – Riva del Garda (TN), 22-25 settembre 2005. “Polymodal activation of stretch-sensitive cation channels in leech neurons” – C. Barsanti, M. Pellegrini, D. Ricci, M. Pellegrino.

Pubblicazioni in extenso

• Pellegrino M, Calabrese B, Menconi MC, Barsanti C, Pellegrini M. (2005) Mechanosensitive cation channels of leech neurons. In Mechanosensitivity in cells and tissues, (Kamkin A, Kiseleva I, eds), Academia, Moscow, pp.337-356. • Barsanti C, Pellegrini M, Ricci D, Pellegrino M. (2006) Effects of intracellular pH

and Ca2+ on the activity of stretch-sensitive cation channels in leech neurons. Eur J Physiol (in press).

• Barsanti C, Pellegrini M, Pellegrino M. (2006) Regulation of the mechanosensitive cation channels by ATP and cAMP in leech neurons. BBA Biomembranes (submitted).

LIST OF ABBREVIATIONS

Ado: adenosine

ADP: adenosine 5’-diphosphate

AMP-PNP: adenosine 5’-(β,γ-imido)triphosphate

ANKTM1: ankyrin transmembrane protein 1

ASIC: acid-sensing ion channel

AP: anterior pagoda

ATP: adenosine 5’-triphosphate

BDNF: brain-derived neurotrophic factor

[Ca2+]i: intracellular calcium concentration

cAMP: adenosine 3’,5’-cyclic monophosphate

DEG: degenerin

ECM: extracellular matrix

EGTA: ethyleneglycol-bis(β-aminoethylether)N,N-tetraacetic acid

ENaC: epithelial Na+ channel

Fq: frequency

HEK: human embryonic kidney

HEPES: N-(2-hydroxyethyl)piperazine-N’-2 ethansulfonic acid

Im: mean current

InsP3: inositol-1,4,5-triphosphate

K2P: two-pore domain potassium channels

MC: multiconductance

mcot: mean channel open time

MS: mechanosensitive

MscL: bacterial mechanosensitive channel with large conductance

MscM: bacterial mechanosensitive channel with “mini” conductance

MscS: bacterial mechanosensitive channel with small conductance

NAD: nicotinamide adenine dinucleotide

NMDA: N-methyl-D-aspartate

nomp: no mechanoreceptor potential

pHi: intracellular pH

PKA: cAMP-dependent protein kinase

PLC: phospholipase C

SA: stretch-activated

SAC: stretch-activated cation channel

SEM: standard error of means

SI: stretch-inactivated

SL: spike-like

SOC: store-operated channel

TM: transmembrane

TRAAK: TWIK-related arachidonic acid-activated K+ channel

TREK: TWIK-related K+ channel

TRP: transient receptor potential

TRPC: “canonical” subfamily of TRP channels

TRPM: “melastatin” subfamily of TRP channels

TRPV: “vanilloid” subfamily of TRP channels

ABSTRACT

Mechanosensitivity is a property shared by ion channels belonging to distinct molecular families, exhibiting a dissimilar structure and different properties concerning ion selectivity, conductance and sensitivity to other physical and chemical stimuli. Stretch-activated (SA) channels are expressed in a wide variety of cell types, including neurons; their specific functions have been clearly demonstrated in sensory reception, such as hearing, touch and osmosensation.

Stretch-sensitive cation channels (SACs) with large conductance have been also identified by means of patch-clamp techniques in membrane patches of non-specialized leech neurons. These channels display high permeability to calcium and their pharmacological features are similar to those of mechanogated channels in vertebrate hair cells. Channel activation has been reported to occur by application of slow negative pressure to the patch pipette or by perfusion with hypotonic solutions.

Two distinct activity modes, differing in their kinetics and conductance, were observed: the first mode, denoted as spike-like (SL), was mainly displayed in patches excised from freshly naked quiescent cell bodies, while the second, called multiconductance (MC) mode, was commonly found in cell bodies and growth cones of cultured neurons.

Gentamicin, a non-specific blocker of mechanosensitive channels, increased neurite outgrowth in culture, suggesting that leech SACs may be involved in axon growth. A gentamicin-sensitive contribution to the calcium increase induced by hypotonic swelling was also revealed by fura-2 imaging.

The present work was initially aimed to provide a better description of the mechanical responses to rapid step pressure changes: the low threshold of activation and the brief opening latency that we observed indicated that these channels could act as rapidmechanotransducers.

Single-channel recordings were then used to study the effects of various intracellular factors on inside-out membranes displaying SAC activity in the two modes. The results show that SL and MC mode can be differently affected by changes in intracellular pH and calcium concentration, as well as by perfusion of the internal face with adenosine nucleotides.

In the absence of mechanical stimulation, intracellular acidification from pH 7.2 to pH 5.5 reversibly increased mean channel open time and opening frequency in the SL mode, while reducing the single channel conductance. Channels in the MC mode were activated by a pHi lowered from 7.2 to 6.2, but

were instead inhibited at pHi 5.5; in this case, the single channel conductance

was unmodified by pH changes.

Unlike MC mode, SL activity was greatly influenced by intracellular free Ca2+ levels. SL channels were activated by calcium concentrations ranging from 1 to 10 µM; this is consistent with a role for these channels in the amplification of Ca2+ signals in leech neurons.

On the contrary, internal perfusion of excised patches with ATP determined a robust and reversible increase in MC channel activity, while the SL mode of expression was unaffected.

In the MC mode, the ATP-induced activation typically exhibited a latency of few seconds and was reported to be dose-dependent, from concentrations as low as 1 µM, up to millimolar levels. This effect was observed at different values of intracellular pH and Ca2+.

Control experiments, carried out by using the non-hydrolysable ATP analog AMP-PNP, ATP without Mg2+ or ADP showed that these nucleotides also effectively enhanced the MC current, with kinetics comparable to that produced by ATP. These findings indicate a direct interaction of the stimulating nucleotides with the channel or with associated molecules. Adenosine itself could reproduce the effect of its nucleotides.

At negative membrane potentials, both ATP and adenosine activated the channel. Moreover, ATP but not adenosine induced a flickering block.

Finally, addition of cyclic AMP, during maximal ATP activation, was able to completely and reversibly inhibit the channel within minutes. This modulation is hypothesized to be mediated by a cAMP-dependent protein kinase, since cAMP alone only induced a weak and rapid channel activation, without inhibitory effects.

In summary, stretch-activated cation channels of leech neurons, like other mechanosensitive channels, display a polymodal regulation, and thus can function as integrators of different environmental signals. These properties, together with their expression in the growth cones and their permeability to calcium, suggest the participation of these channels in the Ca2+ oscillations associated with neuronal growth.

RIASSUNTO

Canali ionici attivati dalla distensione della membrana sono stati descritti in un’ampia varietà di tipi cellulari. Riuniti nel loro complesso sotto la sigla di SA (“stretch-activated”), questi canali appartengono però a distinte famiglie molecolari, ed al di là della comune caratteristica di meccanosensibilità, esibiscono differenti proprietà di selettività ionica, conduttanza e sensibilità a stimoli non meccanici. Alcune loro funzioni specifiche sono state chiaramente dimostrate nella recezione sensoriale, mentre nelle membrane neuronali non specializzate il loro ruolo rimane ancora largamente da definire.

Canali cationici meccanosensibili (SACs) sono stati identificati con tecniche di patch-clamp nelle membrane di neuroni centrali di Hirudo

medicinalis, non specializzati per la meccanorecezione. Questi canali

presentano un’alta permeabilità al Ca2+ ed hanno proprietà farmacologiche simili a quelle dei canali meccanosensibili tipici delle cellule ciliate dell’orecchio dei Vertebrati. L’attivazione può essere ottenuta mediante applicazione di pressione negativa alla pipetta di registrazione o inducendo la distensione della membrana cellulare mediante perfusione con soluzione ipotonica.

Sono stati osservati due diversi modi di attività dei canali, con distinte proprietà cinetiche e di conduttanza: l’attività denominata “spike-like” (SL) è prevalente nei corpi cellulari di neuroni quiescenti, mentre il modo definito “multiconductance” (MC) è maggiormente espresso nella membrana somatica e dei coni di crescita di neuroni in coltura.

In cellule in coltura il trattamento con gentamicina, un bloccante aspecifico dei canali meccanosensibili, aumenta la lunghezza delle

arborizzazioni, indicando un possibile coinvolgimento di questi canali nei fenomeni di estensione assonale. I SACs contribuiscono inoltre agli aumenti del Ca2+ intracellulare in neuroni stimolati con perfusione ipotonica.

Il presente lavoro ha avuto innanzitutto lo scopo di ottenere una migliore caratterizzazione dell’attivazione meccanica dei canali in risposta a variazioni a gradino della pressione applicata al tassello di membrana: la bassa soglia di attivazione e la breve latenza di apertura osservate sono compatibili con un ruolo di questi canali come meccano-trasduttori rapidi.

Esperimenti di registrazione della corrente di singolo canale sono stati poi utilizzati per studiare l’effetto di vari fattori intracellulari sull’attività dei SACs nei due modi di espressione, perfondendo la faccia citoplasmatica di tasselli nella configurazione “inside-out” con soluzioni di diversa composizione. In particolare, sono state prese in esame le variazioni di pHi e della

concentrazione intracellulare di Ca2+, le azioni intracellulari di ATP e adenosina, ed infine l’applicazione di AMP ciclico.

L’acidificazione da pH 7.2 a 5.5 era in grado di attivare reversibilmente i canali nella modalità SL, con aumento della frequenza e della durata delle aperture, mentre induceva contemporaneamente riduzione della conduttanza di singolo canale. I canali nel modo MC erano attivati in seguito all’abbassamento del pH da 7.2 a 6.2, ma venivano invece inibiti a pH 5.5, senza modificazioni di conduttanza.

L’attività di tipo SL, ma non quella MC, era fortemente influenzata dalla concentrazione di Ca2+ libero intracellulare. Elevazioni dei livelli di Ca2+ nella scala da 1 a 10 µM erano efficaci nel produrre attivazione dei canali SL, suggerendo un loro ruolo nell’amplificazione dei segnali Ca2+.

Al contrario, la perfusione della faccia interna dei tasselli con ATP induceva un massiccio e reversibile aumento dell’attività dei canali MC, senza produrre alcun effetto sul modo di attività SL.

Sui canali di tipo MC l’attivazione indotta da ATP si manifestava tipicamente con latenze brevi, dell’ordine di pochi secondi. L’effetto era dose-dipendente in un intervallo di concentrazioni da 1 µM a 1 mM ed era osservabile per diversi valori di pHi e [Ca2+]i.

AMP-PNP, un analogo non idrolizzabile dell’ATP, ATP in assenza di Mg2+ e ADP erano in grado di aumentare l’attività MC in modo paragonabile all’ATP. La stessa adenosina poteva riprodurre l’effetto osservato con i suoi nucleotidi.

ATP e adenosina attivavano il canale anche a potenziali di membrana iperpolarizzanti; in questo caso un blocco di tipo “flickering”, forse legato alla presenza delle cariche negative nella molecola, era contemporaneamente indotto dalla perfusione con ATP, mentre era assente con adenosina.

Durante l’attivazione da ATP, l’aggiunta temporanea di cAMP era in grado di inibire completamente e reversibilmente il canale. L’effetto insorgeva con cinetica lenta, dell’ordine di alcuni minuti, facendo supporre l’intervento di una proteina chinasi cAMP-dipendente.

In conclusione, i risultati ottenuti evidenziano come i canali cationici meccanosensibili dei neuroni di Hirudo medicinalis siano soggetti a regolazione polimodale e possano pertanto integrare segnali ambientali di diversa natura. Queste proprietà, insieme con l’espressione dei canali nei coni di crescita e la loro elevata permeabilità al Ca2+, ne fanno dei potenziali candidati alla genesi e alla regolazione delle oscillazioni di Ca2+ associate ai processi di crescita neuronale.

INDEX

Introduction

p.1

1.

General features of mechanotransduction in living cells

1

3

The pharmacology of mechanogated channels ….………...…………..

6

Touch sensation in C. elegans ….………...…………..……… 9

Drosophila mechanoreceptors ………...………..……… 11

Vertebrate mechanotransduction: the hair cells in the inner ear …….….. 14

Mechanosensitive channels in non-specialized systems ……….…………

17

Ca2+ transients in growth cone migration: a possible role for stretch-activated channels? ……….

19

2.

Molecular diversity of mechanosensitive ion channels

21

Mechanosensitive channels in prokaryotic cells………..…………... 23

The superfamily of DEG/ENaC channels ………..…………...

25

Mechanosensitive K+ channels in the 2P-domain family …………..………

28

The Transient Receptor Potential (TRP) family of ion channels …….…….

31

TRP channels in mechanosensation ………..

35

TRP channels in growth cone guidance ………..

39

3.

The experimental model

40

Materials and methods

p.52

Isolation of leech segmental ganglia ………...

52

Electrophysiological recordings ………...………..

52

Data processing and statistical analysis ……….………...

54

Results

p.56

56

Effects of intracellular acidosis ……….…………..………...

58

Effects of changes in intracellular Ca2+ concentration ………

62

Effects of intracellular application of adenosine nucleotides ………

64

Discussion

p.73

INTRODUCTION

1.

GENERAL FEATURES OF

MECHANOTRANSDUCTION IN LIVING CELLS

Cells experience a wide variety of mechanical forces, including gravity, tension, compression, fluid shear stress and osmotic swelling. Although physical stimuli are potentially damaging for cell membranes, they also represent an important source of information about the characteristics of the extracellular milieu: therefore, mechanical signals, integrated with other environmental input, can exercise their influence on several biological processes, such as cell survival and proliferation, differentiation, cytoskeletal reorganization, metabolism and gene expression (Chen et al., 1997; Ingber, 1997; Iqbal & Zaidi, 2005). Hence, the necessity for living cells, since evolution began, to develop molecular mechanisms able to detect mechanical deformations acting upon the plasma membrane, and transduce them into chemical or electrical messages(Hamill & Martinac, 2001).

Mechanosensitivity represents one of the oldest sensory functions to evolve in living organisms and it is probably a universal property of cells. Touch, hearing, proprioception, baroreception, cell volume regulation, tissue growth and remodeling are some examples of involvement of mechanosensation processes in cell physiology. Specific cells are selectively exposed to specific forms of mechanical stimulations: endothelial cells in blood vessels are exposed to shear stress and stretching forces; bone cells and chondrocytes are subjected to tension and compression produced by gravity and muscular

activities; in cardiac muscle fibers, mechanical stress can regulate the rate of the heart’s electrical activity; in the course of development, mechanical forces generated at the level of cell-cell interactions and from adhesion to specific substrates, during cell guidance and migration, are crucial for the acquisition of a correct tissue morphology.

Moreover, all external stimuli act on top of various internally generated forces, arising from cytoskeletal rearrangements and activity of molecular motors, that contribute to determining cell shape, growth, motility and adhesion

(Ingber, 1997).

Multiple and parallel signalling pathways are probably required to monitor and respond selectively to these different forces. The complex mechanisms responsible for the conversion of mechanical forces into electrical and chemical signals in cells are indicated with the term “mechanotransduction”.

Among the molecules hypothesized to function as primary

mechanotransducers, cell adhesion molecules such as integrins are implicated to a greater extent in slow processes such as tissue remodeling (Katsumi et al., 2004). Integrins are the main transmembrane receptors connecting the cytoskeleton to the extracellular matrix (ECM) in focal adhesion complexes; they can transmit bidirectionally mechanical stresses across the plasma membrane and participate in the activation of intracellular signalling pathways, for example those involved in the control of cell proliferation.

Rapid reactions to applied forces are instead generally attributed to mechanosensitive ion channels. The mechanical perturbation of membrane or associated elements changes the conductive state of mechanosensitive channels, leading to depolarization or hyperpolarization, and to a cascade of intracellular events.

T

First described in embryonic chick skeletal muscle cells (Guharay & Sachs, 1984), mechanosensitive (MS) channels have been identified by electrophysiological techniques in a variety of cell types and have been shown to be expressed in all the living organisms (Morris, 1990; Sachs, 1992; Sackin, 1995). Mechanically-activated currents can be recorded not only in specialized sensory cells, but also in many types of non-sensory somatic cells (Sachs & Morris, 1998) and even in microorganisms. Their ubiquity has led to the idea that mechanogated channels could play a critical role in some general functions, such as cell volume regulation. MS channels may have evolved as cellular osmoregulators, sensitive to changes in tension induced by hypotonic swelling in the membranes of primordial cells, and could have been employed later in the control of cell size and volume as well as in more specialized forms of mechanotransduction. Nevertheless, if the role of MS channels in specialized mechanoreceptors is well-known, their function in many non-sensory cells still remains unclear.

Mechanosensitive channels are grouped together in a common category according to their shared property of mechanical gating. However, they cannot be considered as an homogenous class: on the contrary, they display a wide range of ion selectivities (cation, anion, K+ or non-selective) and values of single channel conductances ranging from tens to thousands of pS (Hamill & McBride, 1996).

In terms of gating, two types of mechanosensitive channels are described, stretch-activated (SA) and stretch-inactivated (SI) ion channels, depending upon whether they are opened or closed by mechanical stimulation,

respectively (Sachs & Morris, 1998). Stretch-activated channels, and particularly those permeable to cations (SACs), have so far received the most attention.

Two models are currently used to describe the mechanism of opening following mechanical stimulations: the “bilayer model” and the “tethered model”, according to whether membrane tension is conveyed to the channel via the surrounding lipid bilayer or by elements anchored to the cytoskeletal and/or ECM (Hamill & Martinac, 2001).

In the bilayer model, first proposed for bacterial mechanosensitive channels, tension in the lipid bilayer is sufficient to gate the channel directly. Purified microbial MS channels, such as MscL and MscS, retain mechanosensitivity when reconstituted into liposomes (Sukharev et al., 1993; Perozo & Rees, 2003). Examples of stretch-activated channels which preserve their mechanical activation in cytoskeleton-deficient membranes can also be found in vertebrates, as shown for SACs in Xenopus oocytes (Zhang et al., 2000a; Maroto et al., 2005).

Instead, the tethered model requires direct connections between mechanosensitive channels and cytoskeletal or extracellular matrix proteins

(Gillespie & Walker, 2001). This model best explains the gating of stretch-activated channels in more specialized structures, such as hair cells, but it has been generally applied to many MS channels in eukaryotic cell types. However, recent advances in the identification of the molecular components of the mechanotransduction apparatus in vertebrate and invertebrate models have questioned whether tethering to rigid elements necessarily implies force transmission, and has led some authors to stress the possibility that force detection mainly occurs at the channel-lipid interface (Kung, 2005).

At any rate, the state of the cytoskeleton and the membrane domain in which the channels are inserted appears to exercise an important influence on their responses to mechanical stimuli. Although SACs exhibit an intrinsic mechanical sensitivity, as shown by their prompt activation in membrane blebs lacking a cortical cytoskeleton, extrinsic factors such as membrane infolding and sub-membrane shock absorber cytoskeletal structures can exercise effects of mechanoprotection, by reducing the fraction of mechanical energy available for channel opening (Ko & McCulloch, 2000).

Cytoprotective adaptations involve a coordinated interaction between the cell membrane and the associated cytoskeleton. For example, cortical actin filaments have been implicated, in the regulation of stretch-activated calcium-permeable channels in fibroblasts, since sustained force promotes subcortical actin assembly and this assembly response “desensitizes” channels to repeated force applications (Glogauer et al., 1997).

In general, treatment with agents that induce depolymerization of actin microfilaments, such as cytochalasins, has been shown to increase the stretch sensitivity of mechanogated channels, as observed in chick muscle and snail neurons (Guharay & Sachs, 1984; Small & Morris, 1994; Wan et al., 1999).

Another potentially mechanoprotective protein is dystrophin, which is one of the structural proteins that underlie and support the sarcolemma. The dystrophic (mdx) mouse is characterized by the absence of dystrophin and is utilized as an animal model of Duchenne muscular dystrophy. Stretch-activated Ca2+-permeable channels display a higher open probability in mdx muscle, so indicating that the absence of dystrophin could contribute to the elevation of [Ca2+]i in these cells(Allen et al., 2005).

An additional function of the cortical cytoskeleton may be to organize the bilayer into local domains by offering a structural support to the folding of large excesses of membrane into microvilli or caveolae (van Deurs et al., 2003). This strategy allows cells to swell without increasing the total bilayer area, and could explain why mechanosensitive currents are not always revealed in whole-cell recordings but only in excised patches (Zhang & Hamill, 2000). Furthermore, membrane compartmentalization into microdomains permits the confinement of ion channels to be locally modulated by the signalling pathways (Lockwich et al., 2000). Stretch-sensitive channels could thus exhibit a different mechano-susceptibility as a consequence of facing different microenvironmental conditions.

THE PHARMACOLOGY OF MECHANOGATED CHANNELS

The molecular identification of stretch-sensitive channels has been slowed till recent years, mainly by the lack of specific toxins. Indeed, despite the discovery of various compounds that clearly block MS channels, none of these drugs appears to be selective, perhaps with the only exception of a peptide isolated from a type of spider venom (Suchyna et al., 2000). So far, the most frequently utilized classes of blockers are represented by amiloride and its analogs, by the antibiotics of the aminoglycoside family and by the lanthanide ion gadolinium. The effects have been studied mainly on stretch-activated cation channels; each class of substances seems to involve a different mechanism of action (Hamill & McBride, 1996).

Extracellular amiloride has been used to block mechanosensitive channels in cochlear hair cells, SACs in Xenopus oocytes, and K+-selective SA channels in molluscan neurons (Sachs & Morris, 1998). The blocking mechanism

is best characterized in SACs from Xenopus oocytes: it exhibits a slight voltage-sensitivity that probably arises from a voltage-dependent conformational change of the channel, resulting in the exposure of two amiloride binding sites. At the single channel level, amiloride induces brief interruptions in the inward current recorded at negative membrane potential, that increase in frequency with the drug concentration (“flickery block”). Amiloride is clearly not selective, since it also blocks epithelial Na+ channels, several voltage-gated channels and even some transporters, including the Ca2+/Na+ exchanger and the Na+/H+ exchanger.

Antibiotics of the aminoglycoside family, such as gentamicin, neomycin and streptomycin, also block mechanosensitive cation channels in hair cells and chick skeletal muscle. Chronic exposure to aminoglycosides causes hair cell death, resulting in significant hearing impairment. Aminoglycosides induce a complete voltage-dependent block. They also block voltage-gated Ca2+ channels and Ca2+-activated K+ channels, although at a higher concentration.

The most common blocker currently used for mechanosensitive channels is gadolinium, a lanthanide trivalent ion that causes a complete and voltage-independent block. Gd3+, along with other lanthanides, has been shown to exhibit strong interactions with lipid bilayers and could exercise its actionnot on the channel protein, but on the surrounding lipids. Although Gd3+ is effective at blocking MS channels in plants, bacteria and in a variety of animal cells at concentrations below 100 µM, it blocks several types of voltage-gated channels as well as voltage-independent leak channels.

Finally, a peptide blocker that seems to be specifically effective for a number of stretch-activated cation channels has been identified in the venom of the spider Grammostula spatulata (Suchyna et al., 2000). Even this toxin, called

GsMTx-4, is hypothesized to enter the bilayer to affect the channel’s surrounding environment. However, its use as a pharmacological tool to test the involvement of SACs in physiological processes is still not very widespread.

MODELS OF SPECIALIZED MECHANOTRANSDUCTION

It is worth describing specialized mechanotrasduction in detail in order to extract some general rules of assembly of the molecules involved in this function.

Mechanosensory transduction underlies a wide range of senses, including proprioception, touch, balance and hearing. Accordingly, all living organisms have developed specific structures, suitable for mediating the reception of mechanical forces in each sensory system. Touch receptors exist in a variety of forms, ranging from free nerve fibers to capsulated structures. Auditory receptors have a specialized apical surface, with actin-based stereocilia. However, despite a different organization, mechanosensory receptors are thought to rely on similar transduction mechanisms. In all these systems, the key element is a mechanically gated ion channel that transduces the forces acting upon the plasma membrane into changes of excitability (Nicolson, 2005).

General features of mechanosensory systems are speed and sensitivity. Speed requires that mechanical forces be conveyed directly to the transduction channel, without intervention of second messengers. Sensitivity requires that the maximal amount of stimulus energy be directed to the transduction channel.

A general model for the mechanotransduction apparatus is based on a 5-component transmembrane system consisting of an ion channel, tethered via intra- and extracellular links to internal and external anchors (Fig.1) (Gillespie & Walker, 2001). The deflection of the external components relative to the internal ones changes the tension in all the elements of the system, and the channel responds by changing its open probability. According to the specific sensory apparatus, an effective stimulus

could take the form of a distortion of the skin, oscillation of a hair cell’s hair bundle, or vibration of a fly’s bristle.

Genetic approaches had been widely used in the search for components of the transduction apparatus in mechanosensory neurons, both in invertebrates and vertebrates. The most commonly used invertebrate models are the nematode Caenorhabditis elegans and the fruitfly Drosophila melanogaster.

TOUCH SENSATION IN C. ELEGANS

In nematodes, sensitivity to body touch is mediated by six touch-receptor neurons, characterized by microtubule-filled processes, enveloped by a mantle of extracellular material and extending along the animal’s sides (Syntichaki &

Tavernarakis, 2004). A distinct class of ciliated nociceptive receptor cells is instead implicated in the detection of mechanical stimuli, osmotic pressure changes and various odorants in the nose region (Mutai & Heller, 2003).

The “gentle body touch response”, that is the withdrawal reaction to a gentle mechanical stimulus delivered transversely along the body wall, is a well-characterized behaviour in these worms. C. elegans mutants, defective in gentle-touch sensitivity

(“mec” mutants) were selected through mutagenesis screening and this approach led to the identification of several molecules required for touch sensation (Chalfie & Au, 1989). Although some of these gene products are certainly involved in the development of mechanosensory touch neurons, other “mec” genes were reported to encode for components of the cytoskeleton, the extracellular matrix or linker proteins, and seem to constitute a transduction machinery, which also comprehends some channel molecules

(Fig.2) (Gillespie & Walker, 2001). Consistent with the “tethered model” for force transmission, extracellular matrix components such as MEC-1, MEC-9 (similar to epidermal growth factor) and MEC-5 (a collagen-like molecule) are thought to convey the mechanical stimulus to the channel proteins MEC-4 and MEC-10. The complex includes an additional subunit, MEC-6, that binds to MEC-2, a stomatin-like molecule which is supposed to serve as an intracellular anchor and to interact with the microtubule cytoskeleton. The genes mec-7 and mec-12 were in fact identified as respectively encoding β- and α-tubulin.

This general model is however still hypothetical, as well as the function proposed for each component in the apparatus. While null mutations in mec-4, mec-6 and mec-2 genes specifically abolish mechanotransduction in body touch neurons, the transduction current is only attenuated in mec-7 mutants (O’Hagan et al., 2005). According to recent analysis, matrix and cytoskeletal elements could have rather a structural role in the localization of transduction channels into punctate clusters, and not to be directly connected with the mechanical gating (Kung, 2005).

The channel proteins identified through genetic screening, MEC-4 and MEC-10, were found to belong to the family of degenerins (DEGs), which also includes other members in the nematode genome, such as unc-8, unc-105 and deg-1, and encodes ion channels related to the vertebrate epithelial sodium channels (ENaCs) (Welsh et al., 2001).

DROSOPHILA MECHANORECEPTORS

Drosophila is certainly one of the favourite models used for studying mechanosensation due to the possibility of combining genetic analysis with electrophysiological recordings from mechanoreceptor neurons, whereas registering in

vivo electrical activity from nematode neurons has proven to be difficult (Goodman, 2004; Ronan & Gillespie, 2005).

The Drosophila mechanosensory system comprises two distinct sets of mechanoreceptors, each exhibiting different general structures. Type Ι sensory organs consist of one to three bipolar neurons, each bearing a single ciliated sensory dendrite, supported by specialized accessory cells; type ΙΙ mechanoreceptors are instead single non-ciliated multidendritic neurons, unaccompanied by accessory cells (Gillespie & Walker, 2001).

Type Ι mechanoreceptors include external sense organs, such as bristle mechanoreceptors, and internal proprioceptors, such as chordotonal organs and Johnston’s organ, which constitutes the fly’s antennal hearing apparatus (Jarman, 2002). Developmental and comparative studies suggest a common evolutionary origin for this class of sensory receptors.

Sensory bristles are particularly suitable for physiological manipulations in the intact animal. Each bristle consists of a hollow hair shaft whose base impinges on the dendritic tip of a bipolar sensory neuron. The shaft acts as a lever arm in which deflections of the external bristle compress the dendritic tip and gate the transduction channels (Fig.3). By removing the tip of the hollow sensory bristle it is possible to gain electrical access to the mechanoreceptor neuron and thus, by placing a recording electrode over its end, to monitor the transduction currents in response to calibrated mechanical stimuli (Walker et al., 2000).

Analogously to the C. elegans screening, a genetic approach was carried out in Drosophila to search for mutants defective in the mechanosensory functions (Kernan et al., 1994). The screening identified a number of larval touch-insensitive mutants that developed into uncoordinated and often deaf adults, as evidence that evolution has conserved similar transduction mechanisms in different mechanosensory modalities in the fly (Eberl et al., 2000). These mutants showed either a reduced mechanoreceptor potential (“remp” mutants) or no mechanoreceptor potential at all (“nomp” mutants) in

bristle neurons. In particular, two genes (nompA and nompC) were characterized, encoding respectively an extracellular matrix protein and an ion channel. NompA is a modular protein secreted by the supporting cells, which is supposed to serve as an extracellular mechanical link between the sensory dendrites and the associated cuticular structures. The nompC gene product is a cation channel belonging to the transient receptor potential (TRP) family and is also known as TRPN1 (Walker et al., 2000). The presence of an extraordinarily high number of ankyrin repeats, a common protein-protein interaction motif (Sedgwick & Smerdon, 1999), in the NompC intracellular N-terminal domain may support the hypothesis of a connection between the transduction machinery and the underlying cytoskeleton.

Electrophysiological evidence is consistent with a function of NompC as a mechanosensory transduction channel: loss-of-function nompC mutants show loss of almost all the transduction current, whereas point mutations can change the adaptation

profile (Walker et al., 2000). However, although this channel accounts for about 90% of the mechanically-activated current, a residual response observed in nompC mutants suggests the presence of an additional channel sensitive to mechanical stimulations. NompC appears to be selectively expressed in type Ι mechanosensory organs. An orthologue of NompC/TRPN1 can also be found in some ciliated mechanoreceptor neurons in C. elegans, as well as in zebrafish sensory hair cells (Sidi et al., 2003). Another TRP-like channel, called Nanchung, seems instead to be required as the primary mechanotransducer in chordotonal organs (Kim et al., 2003).

Although less studied, type II non-ciliated neurons of flies probably use a distinct mechanism of mechanotransduction. Some are known to express members of the DEG/ENaC family of Na+ channels, such as Pickpocket (PPK) (Adams et al., 1998), analogously to the non-ciliated touch-receptor neurons in nematodes. Thus, two separate kinds of specialized mechanoreception, relying on a different cell organization and molecular basis, may have evolved in invertebrates, both of which have been conserved in distinct cell types.

VERTEBRATE MECHANOTRANSDUCTION: THE HAIR CELLS IN THE INNER EAR

Hair cells, the mechanoreceptors of the inner ear, transduce auditory and vestibular stimuli into electrical signals. Mechanical signals are detected as deflections of the hair bundle, a specialized structure on the apical surface, consisting of a cluster of tens to hundreds of actin-filled stereocilia and a single axonemal kinocilium. Adjacent stereocilia are connected with each other through extracellular elastic filaments, the tip links, which are probably involved in the transmission of forces, and whose major component has been recently identified as cadherin 23 (Gillespie et al., 2005). An excitatory deflection of the hair bundle directly opens cationic transduction channels, located at the top of stereocilia. The resulting inward current depolarizes the cell membrane and subsequently triggers neurotransmitter release. On the contrary,

inhibitory deflections close the transduction channels and hyperpolarize the cell (Fig.4) (Gillespie & Walker, 2001; Nicolson, 2005).

The transduction channel of hair cells has some peculiar properties that distinguish it from other channels (Strassmaier & Gillespie, 2002): noticeably, it is a non-selective cation channel, with a large conductance (about 100 pS) and a substantial permeability to calcium. With respect to pharmacology, the channel is blocked by relatively low concentrations of aminoglycoside antibiotics, by amiloride and its derivatives, and by lanthanide ions, particularly Gd3+.

Transduction channels open within microseconds after the application of a mechanical stimulus (Corey & Hudspeth, 1983); this extremely rapid response precludes the involvement of a second messenger system. Like other sensory receptors, hair cells respond to sustained stimuli by adapting. Two distinct forms of adaptation are described. Fast adaptation requires a millisecond or less and is a calcium-dependent process: Ca2+ enters the cell through the transduction channel and probably binds to a site near the channel itself, determining its closure, with a typical negative feed-back mechanism. Slow adaptation, which requires tens to hundreds of milliseconds, is instead mediated by a molecular motor, probably an unconventional isoform of myosin, known as myosin-1c or myosin Ιβ, located near stereociliary tips (Gillespie & Cyr, 2004). Slow adaptation is regulated by Ca2+ concentration too, since interaction of myosin-1c with hair-cell receptors occurs through calmodulin-binding domains (Cyr et al., 2002).

Although the biophysical features of mechanical transduction in hair cells were first described long ago, the molecular identity of the transduction channel has only recently begun to be clarified. The difficulty in accessing the auditory organs, the paucity of sensory hair cells, the small number of channels, together with the lack of highly specific toxins, all may have contributed to its delayed identification. According to recent studies, the most plausible candidate for the mechanosensitive transduction channel in vertebrate hair cells is TRPA1, another member of the transient receptor potential family (Corey et al., 2004; Nagata et al., 2005).

Although TRPA1 does not exhibit a great sequence homology with TRPN1 (NompC), the supposed mechano-gated channel in fly sensory bristles, it is interesting that both channels, uniquely among the members of the TRP superfamily, share the unusually high number of ankyrin repeats. Moreover, the mechanoelectrical responses of Drosophila bristles and vertebrate hair cells have similar features: even the fly mechanoreceptor currents arise with short latencies (≈ 200 µs), indicating a direct opening of the transduction channel, and display adaptation (Walker et al., 2000).

Other similarities between the Drosophila sensory bristle and the vertebrate hair-cell transduction systems concern the developmental pathway, the general structure of ciliated receptor neurons (although with some differences: cilia have an actin skeleton in hair cells, whereas they have a microtubular one in Drosophila sensory organs), the directional sensitivity to displacements, the high extracellular K+ concentration (Nicolson, 2005). Thus, it has been speculated that vertebrate hair cells and ciliated mechanoreceptors of invertebrates may be ontogenetically related.

M

-

In addition to their obvious function in specialized sensory reception, mechanosensitive ion channels have been identified by single-channel recordings in many non-specialized cells, where apparently specific macroscopic cellular arrangements are not present (Sachs & Morris, 1998).

Indeed, for technical reasons, related to the inaccessibility of tissues and to the fragility of the ciliated endings of specialized mechanoreceptors, mechano-gated currents were first characterized in non-sensory cell types, such as skeletal muscle fibers (Guharay & Sachs, 1984), endothelial cells (Lansman et al., 1987), fibroblasts (Stockbridge & French, 1988), and Xenopus oocytes (Yang & Sachs, 1990), where their physiological role remains mostly to be determined.

An involvement of mechanosensitive channels in osmosensation and cell volume regulation has been postulated in different preparations (Christensen, 1987; Chen et al., 1996). In magnocellular neuron cells, a subpopulation of hypothalamic neurons that are responsible for detection of blood osmolarity, stretch-inactivated cation channels are demonstrated to transduce osmotically-evoked volume changes into variations of membrane potential, that control, in turn, the release of antidiuretic hormone (Oliet & Bourque, 1993).

However, the widespread diffusion of stretch-sensitive channels may be consistent with a role in other general functions. As already pointed out, among mechanosensitive channels, those permeant to calcium have aroused the greatest interest: stretch-activated Ca2+ influx can in fact provide a direct connection between mechanical stimulation of the plasma membrane and activation of intracellular biochemical responses. Mechanical forces trigger elevation in [Ca2+]i in many cells, including cardiac and smooth muscle cells,

fibroblasts, endothelial cells and neurons (Davis et al., 1992; Sigurdson et al., 1992; Naruse & Sokabe, 1993; Glogauer et al., 1995; Sharma et al., 1995).

Changes in intracellular calcium are implicated in a wide variety of cellular processes, and particularly in molecular events that are essential for cell motility. Ca2+ is instead required for the production of the contractile forces in actomyosin systems, for the regulation of the actin cytoskeleton dynamics, for the formation and disassembly of cell-substratum adhesions (Lee et al., 1999). A role for stretch-activated Ca2+-permeable channels in motility processes, during which cell membranes undergo changes in tension, has been indicated by studies on cell migration carried out in epithelial cells and fibroblasts (Lee et al., 1999; Doyle et al., 2004; Munevar et al., 2004). These cells exhibit a mode of movement with distinct phases of extension of the front edge and retraction of the rear edge. Transient increases in intracellular calcium, determined by influx through SA channels, can be visualized when cells become temporarily attached to the substratum and are subjected to mechanical stretching; the subsequent rise in [Ca2+]i is involved in the generation of localized traction

forces and in the detachment of the rear cell margin. Stretch-activated Ca2+ influx is also involved in mediating the orientation response of endothelial cells to cyclic mechanical stress (Naruse et al., 1998).

CA2+ TRANSIENTS IN GROWTH CONE MIGRATION:

A POSSIBLE ROLE FOR STRETCH-ACTIVATED CHANNELS?

Extending neurites of developing or regenerating neurons migrate through complex environments to reach appropriate target regions. The guidance of nerve fibers to their final destination can be considered as a series of short-range projections under the influence of local cues: growth cones detect external cues through receptors on their surface, and then generate intracellular signals that determine the direction in which the axon grows. In addition to diffusible chemotropic or chemorepulsive factors, mechanical interactions between cell and substrate are also important for axon guidance. Mechanical tension can represent a direct stimulus for axon differentiation and extension (Lamoureux et al., 2002) and a linear relation between pulling tension generated by growth-cone advance and net neurite elongation has been demonstrated (Lamoureux et al., 1989). Moreover, the nature of the substrate on which growth cones navigate has a powerful effect on the remodelling of neuronal processes (Gomez et al., 2001).

Cytoplasmic Ca2+ signals have fundamental roles in regulating axon motility and pathfinding (Henley & Poo, 2004). Neuronal growth cones generate periodic spontaneous [Ca2+]i elevations as they migrate in vivo, and the rate of

axon outgrowth is inversely proportional to the frequency of Ca2+ transients, which therefore appear as a natural signalling mechanism to regulate axonal extension (Gomez & Spitzer, 1999). Growth cones producing a high frequency of Ca2+ transients migrate slowly or retract, whereas growth cones generating a low frequency migrate rapidly. In vivo different classes of neurons exhibit a

specific frequency of Ca2+ oscillations, which also depends on the position of growth cones along their pathway: thus, the immediate environment

surrounding the growth cone participates in determining its Ca2+ dynamics. Although Ca2+ release from intracellular stores can mainly account for the amplitude and the amplification of the signal, Ca2+ influx through non-voltage-gated channels of the plasma membrane is required for the genesis of [Ca2+]i

spikes (Gomez et al., 1995).

While global increases in [Ca2+]i can regulate neurite growth, Ca2+ signals

localized asymmetrically to one side of the growth cone are implicated in attractive and repulsive turning responses (Zheng, 2000). Local Ca2+ signals, whose frequency is also substrate-dependent, can propagate back to the growth cone central domain and stimulate more global [Ca2+]i elevations that

promote growth cone turning (Gomez et al., 2001). Even in this case there is evidence for an involvement of non-voltage-gated Ca2+-permeant channels, which are currently in the process of being identified.

The presence of stretch-activated channels in growth cone membranes has been demonstrated in some invertebrate neurons (Sigurdson & Morris, 1989; Pellegrino et al., 1990). On the other hand, mechanically-induced calcium responses, probably mediated by activation of stretch-sensitive cation channels, have been observedin the cell bodies of non-specialized neurons (Sharma et al., 1995; Gotoh & Takahashi, 1999; Viana et al., 2001). Since it is well-established that neurite and growth cone membranes experience large changes in tension during their extension and retraction, the contribution of mechanosensitive channels to these dynamics, although at present not supported by direct evidence, appears to be plausible.

2.

MOLECULAR DIVERSITY OF MECHANOSENSITIVE

ION CHANNELS

Mechanosensitive ion channels, which detect and transduce external mechanical stimuli into electrical signals, are defined by the functional property that their gating is responsive to membrane deformation (Sachs & Morris, 1998). However, from a molecular point of view, there is no common sequence that identifies mechanosensitivity. On the contrary, ion channels whose activity can be affected by mechanical stresses constitute a heterogeneuous group, formed by members belonging to distinct and evolutionarily unrelated families; they exhibit dissimilar molecular structures and different properties of ion selectivity, conductance and sensitivity to non-mechanical stimuli (Sukharev & Anishkin, 2004; Martinac, 2004).

The structural diversity of mechanosensitive channels probably derived from the physiological necessity of cells to be responsive to forces of different natures and orders of magnitude. Thus, mechanoreceptors arose independently several times during the course of evolution, as an adaptation to specific cellular needs. Each time, the problem of detecting a specific force was solved differently by recruiting a member from an existing family of channels. In fact, families that include mechanosensitive channels often also contain members which are not mechanosensitive. This evolvability of channel families could have been favored by the fact that mechanical force is not a ligand, and so does not need a specific binding pocket. On the contrary, according to one point of view, it is possible that the evolution of voltage-gated or ligand-gated channels made it necessary to strengthen certain intramolecular or intermolecular interactions to exclude the influence of mechanical perturbations on channel

opening (Sukharev & Anishkin, 2004). Consistent with this hypothesis, some characterized channels with functions other than mechanosensation, such as the NMDA receptor (Paoletti & Ascher, 1994), the Drosophila Shaker K+ channel

(Gu et al., 2001) or the N-type Ca2+ channel (Calabrese et al., 2002), still retain properties of modulation by membrane stretch.

The wide distribution of mechanosensitive channels in all cell types, together with the fact that they are found in all living organisms, beginning from the simplest unicellular Bacteria and Archaea (Kloda & Martinac, 2001), points toward their early evolutionary origins.

In the following table the main families of channels which comprehend members with a supposed mechanosensory function are illustrated (Tab.1).

Apart from the bacterial mechanosensitive channels, which are the best characterized thanks to protein purification and X-ray crystallography studies, eukaryotic ion channels involved in mechanotransduction have been identified among the superfamily of DEG/ENaC channels (Tavernarakis & Driscoll, 1997), the two-pore (2P)-domain K+ channels (Patel & Honoré, 2001) and several subfamilies of transient receptor potential (TRP) cation channels (Clapham, 2003). A remarkable feature of many of these channels is that they often display properties of activation by a variety of environmental and intracellular factors, which makes them stand out interestingly as polymodal molecular sensors, suitable for the integration of different physical and chemical signals.

MECHANOSENSITIVE CHANNELS IN PROKARYOTIC CELLS

Prokaryotic pressure-sensitive channels were first identified in 1987 in Escherichia coli, by applying the patch-clamp technique to bacterial sphaeroplasts (Martinac et al., 1987). Since then, multiple MS channel activities have been reported in the E. coli inner membrane: according to their single channel properties, they have been classified as MscL (large), MscS (small) and MscM (mini), with conductance values of about 3 nS, 1 nS and 300 pS, respectively (Hamill & Martinac, 2001; Perozo & Rees, 2003).

Microbial MS channels are assumed to function mainly as osmoregulators. Bacteria exposed to a hypoosmotic shock rapidly react by releasing cytoplasmic contents into the surrounding medium. MS channels clearly play a relevant role in the response to osmotic stress and in the protection of prokaryotic cells from lysis, as demonstrated by the fact that E. coli mutants lacking both MscL and MscS activities die when transferred to a medium of low osmolarity (Levina et al., 1999).

There is a relationship between the size of conductance and the activation pressure required for each of these channels: MscM exhibits the lower threshold,

followed by MscS and then by MscL, which opens at membrane tensions near the lytic limit of the bilayer. The different pressure sensitivities of the three channels may indicate that they are activated sequentially to provide a graduation of efflux pathways in response to hypotonic stress.

MS channels may also be involved in sensing the changes in turgor pressure during cell division and growth and participate in the signalling pathways that regulate these processes (Martinac, 2004).

The large-conductance channel MscL, encoded by the gene mscL, is an oligomer of five identical subunits. It does not exhibit any cation/anion selectivity. The three-dimensional structure of the MscL homolog from Mycobacterium tuberculosis (Tb-MscL), in the closed conformation, has been resolved by X-ray crystallography

(Fig.5a) (Chang et al., 1998). Each subunit is composed of two transmembrane segments, TM1 and TM2, interconnected by an extracellular loop, and by cytoplasmic N- and C-terminal domains. The five central TM1 helices form the pore of the channel, while hydrophobic TM2 helices face the lipid environment. MscL is reported to open like the iris of the lens: the TM1 helices tilt with respect to the membrane plane and cause the channel to flatten (Sukharev & Anishkin, 2004).

The small-conductance channel MscS displays a slight preference for anions over cations and is also modulated by voltage. The channel activity is actually the result of two distinct, although similar, channel proteins that originate from two related gene products: MscS, encoded by YggB, and MscK, encoded by KefA. The crystal structure of MscS has been also determined and it appears to be quite different from that of MscL (Bass et al., 2002): the channel is a homoheptamer with a large cytoplasmic region (Fig.5b). Each subunit contains three transmembrane domains: the TM3 helices form the channel pore, while TM1 and TM2 helices are sensors for voltage and membrane tension.

Finally, MscM activity is observed less frequently than the other channels in patched sphaeroplasts; so far the gene responsible has not been identified.

From a structural point of view, to date bacterial channels are the best characterized mechano-gated channels. In spite of this, no close homolog has been discovered among eukaryotes and certain properties of the microbial channels, including their high conductance, largely greater than that of eukaryotic channels, and the requirement of near-lytic activation tensions appear to be unique to prokaryotes.

T

DEG/EN

C

Genetic screening carried out in C. elegans to identify components of its mechanosensory apparatus led to the discovery of the first members of the degenerin/epithelial Na+ channel (DEG/ENaC) family (Tavernarakis & Driscoll, 1997): MEC-4 and MEC-10 may form the core of a multiprotein ion channel

complex that opens in response to mechanical stimulation. Recessive mutations in mec-4 and mec-10 cause the animals to be insensitive to light touch, whereas gain-of-function mutations in these genes, as well as in deg-1, produce swelling-induced degeneration and lysis of the nematode mechanosensory neurons, consistently with a continuous opening of mechanosensitive channels

(Welsh et al., 2001).

Thus, MEC-4, MEC-10 and DEG-1 were grouped in a protein superfamily called degenerins (DEGs). Other members of the DEG family in C. elegans include unc-8 (uncoordinated), expressed in motor neurons and required for normal locomotion (Tavernarakis et al., 1997), and unc-105, expressed in muscle and required for stretch sensitivity (Liu et al., 1996).

DEG/ENaC proteins share a common topology: members have short intracellular N and C termini, two membrane-spanning sequences and a large extracellular loop with 14 conserved cysteine residues, which form intrachain disulfide bonds (Welsh et al., 2001). Individual DEG/ENaC subunits can assemble as homomultimers or heteromultimers to form cation channels. The channels are voltage-insensitive with a permeability of Na+>>K+: some channels also exhibit a limited Ca2+ permeability (Bianchi et al., 2004). Extracellular amiloride inhibits currents at variable effective concentrations.

Although DEG/ENaC proteins are believed to form sensory mechano-transduction channels, until a short time ago the evidence for this role was indirect. In a recent publication, the first whole-cell recordings of mechano-receptor currents obtained from C. elegans touch sensory neurons have been presented (O’Hagan et al., 2005). Inward currents, carried mostly by Na+ and blocked by amiloride, were activated with a maximum latency of 5 ms upon application of mechanical stimuli with a calibrated glass probe; these currents,

which were apparently mediated by MEC-4, rapidly decreased by displaying adaptation, to turn on again upon release of force, similarly to receptor potentials in vertebrate Pacinian corpuscles. In addition, calcium imaging experiments have shown that MEC-4 and the associated protein MEC-2 are required for generating increases in intracellular calcium in response to gentle mechanical stimulation in vivo (Suzuki et al., 2003a).

Because of their critical role in nematode touch-receptor neurons, it has been proposed that members of the DEG/ENaC superfamily function as mechanosensory transduction channels in different cell types. On the basis of sequence similarities, several other amiloride-sensitive Na+ channels have been classified as DEG family members. Among the Drosophila members of this family, Pickpocket is selectively expressed in sensory multidendritic neurons, responsive to touch and changes in body shape, where it localizes to dendritic varicosities, the likely site of mechanotransduction; however, although this protein is a probable ion channel subunit, it appears to be unable to form channels on its own (Adams et al., 1998).

Various members of the DEG/ENaC family have been identified in mammals: they comprise α, β, γ and δ subunits of epithelial Na+ channels, responsible for Na+ reabsorption in many epithelia, and some acid-sensing ion channels (ASICs), such as BNC1, ASIC and DRASIC, which are activated by extracellular acidosis. Some of these transcripts and/or proteins have been detected in specialized cutaneous mechanosensory structures, including Pacinian corpuscles, Meissner corpuscles and Merkel cell-neurite complexes. However, the role of DEG/ENaC proteins in vertebrate mechanosensation is still poorly characterized (Welsh et al., 2001; Lumpkin & Bautista, 2005).

Although some authors showed that the α-subunits of the epithelial Na+ channel may form a stretch-sensitive channel when reconstituted into planar lipid bilayers (Awayda et al., 1995) and that epithelial Na+ channels heterologously expressed in Xenopus oocytes are regulated by shear stress and osmotic pressure (Ji et al., 1998; Satlin et al., 2001; Carattino et al., 2004), no convincing electrophysiological evidence supporting the mechanical activation of these channels in vivo has yet been reported.

M

K

2P-

The two-pore (2P)-domain potassium channel family comprehends leak or background K+ channels which are thought to contribute to the maintainance of the resting membrane potential. Among the K+-selective conductances, K2P channel subunits display a peculiar structure with four transmembrane segments and two P (pore forming) regions disposed in tandem. Since all functional K+ channels are assumed to be multimeric complexes comprising four P-domains, K2P channels are believed to associate as dimers _{(O’Connell et} al., 2002). These channels are apparently insensitive to the most classical K+ channel blockers, such as tetraethylammonium and 4-aminopyridine.

In mammals, several members of this family are found to be highly expressed in the central and peripheral nervous system. The first member to be identified has been TWIK-1 (for “tandem of P domains in a weakly inward-rectifying K+ channel”) (Lesage et al., 1996), but different subgroups of K+ channels of the two-pore-domain family are currently classified (Patel & Honoré, 2001).

In particular, TREK-1 (for “related to TWIK-1”), TREK-2 and TRAAK (for “activated by arachidonic acid”) belong to a subgroup of K2P channels which

can be reversibly gated by membrane stretch (Patel et al., 1998; Maingret et al., 1999a; Bang et al., 2000). All these channels have been cloned and transfected in heterologous systems, where their biophysical, pharmacological and regulatory properties have been extensively investigated in recent years. Great interest has been focused on these channels, which appear to function as polymodal sensors (Fig.6), due to their characteristics of activation in response to many different physical (stretch, cell swelling, heat, voltage) and chemical (intracellular acidosis, polyunsaturated fatty acids, lysophospholipids, volatile general anesthetics) stimuli (Patel & Honoré, 2001).

Single-channel recordings, obtained with the patch-clamp technique in the cell-attached and inside-out configurations, have shown that TREK-1, TREK-2 and TRAAK channel activity can be stimulated by increasing the

mechanical pressure applied to the cell membrane. At the whole-cell level, TREK-1 and TRAAK are also modulated by cellular volume (Patel et al., 1998). The sensitivity to mechanical stretch is enhanced by treatment of cell-attached patches with cytoskeleton disrupting agents, such as colchicine and cytochalasin (Maingret et al., 1999a), suggesting that mechanical forces might be transmitted directly via the lipid bilayer (Chemin et al., 2005). Finally, both TREK-1 and TRAAK are blocked by amiloride and Gd3+ (Maingret et al., 2000).

Lowering intracellular pH shifts the pressure-activation relationship of TREK-1 and TREK-2 towards positive values and ultimately leads to channel opening at atmospheric pressure (Maingret et al., 1999b; Bang et al., 2000); on the contrary, TRAAK channel is not opened by intracellular acidosis but appears to be more sensitive to alkalosis. Deletional analysis demonstrates that the carboxy-terminal region is crucial for the activation of TREK-1 by stretch and intracellular acidosis (Honoré et al., 2002).

Various lipid compounds are effective on this group of channels. In addition to mechanical sensitivity, another feature common to TREK-1, TREK-2 and TRAAK is that they are reversibly opened by long chain polyunsaturated fatty acids, such as arachidonic acid; the activation mechanism is supposed to be direct, by interaction with the channel protein itself or by partitioning into the lipid bilayer (Patel et al., 2001). TREK and TRAAK channels are also opened by lysophospholipids: in this case activation is probably indirect and may involve a cytosolic factor (Maingret et al., 2000). Opening by inhalational anaesthetics indicates that these channels, as well as other members of the K2P family, may contribute to general anaesthesia.

TREK channels are additionally regulated by second messenger pathways. Phosphorylation of a C-terminal serine residue by cAMP-dependent