A possible unifying principle for
mechanosensation
Ching Kung1
Of Aristotle's five senses, we know that sight, smell and much of taste are initiated by ligands binding to G-proteincoupled
receptors; however, the mechanical sensations of touch and hearing remain without a clear understanding of
their molecular basis. Recently, the relevant force-transducing molecules--the mechanosensitive ion channels--have
been identified. Such channel proteins purified from bacteria sense forces from the lipid bilayer in the absence of other
proteins. Recent evidence has shown that lipids are also intimately involved in opening and closing the mechanosensitive
channels of fungal, plant and animal species.
A
ll creatures have mechanical senses: insects hear, and,
when touched, worms twitch and sea anemones contract.
Touching the front of a unicellular paramecium makes it
swim backward; touching its posterior makes it spurt
forward. Plant roots and shoots respond to gravity (gravitropism),
and stems proportion the growth in their height, versus the growth in
their girth, based on the amount of jostling by wind and rain
(thigmomorphogenesis). Besides the ear and the skin, animals have
many other mechanosensors; for example, the circumventricular
organs (for determining systemic osmolarity), baroreceptors
(blood pressure), spindle receptors (muscle stretch), proprioceptors
(limb positions) and so on. Our bones measure stress during periods
of growth or regeneration, and our tongues sense texture and size so
that we don't swallow sand or stones.
Mechanical senses differ from other senses. The molecular bases of
sensing odorants, hormones, neurotransmitters and other dissolved
ligands (solutes) are well understood: the lock-and-key binding of
each ligand to the specific binding pocket of its specific receptor on
the plasma membrane. Much less is known, however, about the
molecules that sense forces such as osmotic force, thirst, touch,
vibration and texture. Many membranes are equipped with mechanosensitive
(MS) ion channels that respond to turgor in proportion
to the surrounding concentration of water (the solvent). Such
channel proteins have been cloned and crystallized from bacteria.
Examination of these proteins by genetic, electric, chemical and
physical means has found that they are able to directly detect and
respond to forces from the lipid bilayer. The study of MS channels in
plants and animals lags behind, partly because their anatomical
complexities resist reductionistic approaches. Nonetheless, recent
findings indicate the involvement of lipids in the gating of MS
channels of the worm, the fly, the frog oocyte and in mammals.
The same types of channel proteins--for example, transient receptorpotential
(TRP) proteins--are now found to sense vibration, touch
and osmotic membrane stretch. The possibility is emerging that force
detection ultimately occurs at the channel­lipid interface. Displacement
at the interface, either by deforming the bilayer or by pulling the
channel or the bilayer with a tether, provides the energetic drive for
the channel to reach the open conformation.
This Review covers many of the principal advances in this field,
and draws on widely diverse research projects: from hair cells to
bacteria. Traditional compartmentalization of research disciplines,
such as the segregation of micro- and neuro-biology, often hampers
dialogue among projects. Yet, life's basic machineries--such as DNA
replication, transcription, translation, the Krebs cycle, electron
transport, and now ion filtration and voltage gating--are found to
be universal, even though they were originally based, in large
measures, on experiments done in microbial systems. This Review
asks whether there is a common physicochemical basis that determines
how channel proteins sense force, which may serve to unite the
varied biological manifestations. Even though touch, hearing and
osmosis are seemingly disparate fields of research, they all deal with
a single physical parameter--force. Regardless of frequency or
duration, a dyne is a dyne.
MS channels allow bacteria to withstand rain
Life is largely aqueous chemistry--we went to Mars looking for
evidence of water and, by inference, life. Cells comprise ,80% water,
and de- or over-hydration can be lethal. Osmotic force, a measure of
water content, is therefore fundamental to cell survival. Even though
water is 55.6 M, it only requires a 10 milli-osmolar difference across
a barrier to produce an osmotic pressure of ,180 mm Hg
(,2.5  105
dyn cm2 2
), which generates a ,12 dyn cm2 1
stretch
force on the surface of a 2-mm-diameter sphere. Surprisingly, the
proteins in bacteria that directly gauge such stretch forces were only
discovered in the last decade, and only by serendipity. When a cell is
caught in the rain, or following laboratory dilution, the inward
diffusion of water through the lipid bilayer--water channels are not
needed--generates a huge turgor that can rise to hundreds of
atmospheres (108
dyn cm2 2
): far beyond what the cell envelope can
withstand. It is known, from work first carried out in the 1950s, that
Escherichia coli bacteria release their osmolites (solutes) upon an
osmotic down-shock (Fig. 1a). For instance, a 1-in-100 aqueous
dilution of the culture drains the cellular 14
C-labelled proline pool by
.95% (ref. 1). Tracing other osmolites, such as K
, lactose and ATP,
gave similar results2
. However, osmolite-depleted bacteria retain
their macromolecules, and begin protein synthesis within minutes
upon return to normal medium1
. The identity of the `emergency
valves' for solute release remained elusive for almost half a century.
Though MscL and MscS (mechanosensitive channel of large or small
unitary conductance) were postulated to be these valves when their
electrical activities were discovered in 1987 (ref. 3), this role was not
proved until 1999 when Booth and co-workers4
showed that DmscL
REVIEWS
1
Laboratory of Molecular Biology and Department of Genetics, University of Wisconsin, 1525 Linden Drive, Madison, Wisconsin 53706, USA.
Vol 436|4 August 2005|doi:10.1038/nature03896
647
 2005 Nature Publishing Group
DmscS double mutants lyse even upon a rather mild down-shock.
MscL and MscS have redundant functions and single mutants have
no clear phenotype, so they were not discovered in the extensive and
intensive genetic dissection of E. coli physiology in the last century,
but were, in fact, identified in a patch-clamp survey (discussed
below). MS channels are just one--albeit the quickest one--of a
bacterium's many defences against de- or over-hydration. The slower
types of these defences are those that are transcriptionally controlled.
Prokaryotic MS channels
MscL and MscS were encountered in the first electrophysiological
survey of bacterial membranes, conducted by Martinac et al.3
.
Voltage-controlled (-clamped) patches of E. coli membranes produced
giant steps in unitary current when the patches were subjected
to suction3
, or when the solution bathing the patch was diluted5
.
These MS-channel activities remain following reconstitution into
artificial liposomes after membrane dissolution or protein purifi-
cation6
(Fig. 1b). The non-selective unitary conductances of MscL
(,3 nS) and MscS (,1 nS) are 10­1,000-fold greater than those of
the more selective channels commonly studied. Such large signals
and unlimited bacterial material allowed the tracing of the channel
activity in chromatographic fractions to a single protein, which, in
turn, led to the discovery of the mscL gene7,8
. The corresponding
small protein has two transmembrane (TM) helices, M1 and M2
(refs 7, 9; Fig. 2, left). Chang et al. resolved the crystal structure of the
Mycobacterium tuberculosis MscL homologue as a homopentamer
with the five M1 domains converging to close the pore at the
cytoplasmic side10
(Fig. 2, centre). In 2001, computational modelling
and cross-linking experiments led Sukharev et al.11
to a model in
which all TM helices recline and rotate, like the iris of a camera, to
open a large pore of some 30 A° in diameter (Fig. 2, right). The
following year, through observations with electron paramagnetic
resonance spectroscopy after site-directed probe attachment,
Martinac and co-workers12,13
concurred with the main premise of
this model. The cloning4
and crystallization14
of the E. coli MscS
protein showed that it is a homoheptamer of three-TM subunits with
seven converging M3 domains: entirely different from MscL. Genetic,
biochemical, biophysical, simulation and other studies on MscL and
MscS are periodically reviewed15­18
.
Forces from lipids gate prokaryotic MS channels. Exercising
ultimate reductionism, mechanosensitivity was found to be intact
when pure MscL protein was placed in bilayers of one12,13
or two7
defined lipids (Fig. 1b). Here, the stretch force detected by the protein
must come from the lipids themselves, since there are no other
components that could contribute. MscL follows a Boltzmann
distribution where the mechanical energy partitions the MscL
protein between its closed and open conformations, with the midpoint
tension (50% of the channel being open) at ,12 dyn cm21
(ref. 19)--a sensitivity presumably tuned for its biological role
(Fig. 1a). (The threshold tension of MscL is much lower; wild-type
MscS (ref. 20) and MscL gain-of-function mutants21,22
have even
lower thresholds.) MscL functions in bilayers made of ordinary lipids
with charged or uncharged heads, saturated or unsaturated tails, and
in various mixtures. Shortening the length of the fatty acid chain
from 20, to 18, to 16 carbons reduces the energy barrier between the
closed and the open state, but does not trigger spontaneous channel
opening12,13
.
The prevailing model of the behaviour of MscL and MscS
addresses the forces within the lipid bilayer itself. Unlike the aqueous
solution, the bilayer is highly anisotropic: having very different
physical properties at different depths. The free-energy reduction
in ordering waters and lipids at the interface is reflected in a large
surface tension between the lipids' polar head groups and the nonpolar
tails. However, pressures nearby serve to balance this tension,
allowing the bilayer to self-assemble into a stable structure. The force
profile of pure lipid bilayers has been calculated by Cantor23,24
(Fig. 3a) and examined with molecular-dynamics simulation25,26
.
(How the protein­lipid interaction may distort the profile at the
interface is unclear, though it is being analysed27
.) The amplitudes of
these forces are in the order of hundreds of dyne cm2 1
(ref. 25);
much larger than the lytic tension (tens of dyne cm2 1
) of the bilayer.
Any protein embedded in the bilayer is subject to these strong,
localized push or pull forces. Altering the force profile by membrane
stretch, or by channel or lipid displacement through a tether (discussed
below), may make it energetically more favourable for the
channel protein to assume a new conformation; for example, the
open state (Fig. 3b). Among the molecular-dynamics simulations
carried out by several groups, Gullingsrud and Schulten26
directed
forces to residues within the MscL protein at the level of the lipid's
glycerol backbone between the head groups and the tails, where
tension is maximal (yellow arrows, Fig. 2). Such a simulation can
reveal steric clashes or structural disintegration, which should not
happen when a channel opens. They traced the positions of the
111,079 atoms of 1 MscL protein, 365 lipids and 22,308 waters, and
found that MscL indeed opens on a 10-ns timescale in an iris-like
manner, similar to the original model11­13
.
Besides external forces, the composition of the lipid bilayer itself
can affect its internal forces. Adding chemically unrelated cationic
Figure 1 | Bacterial channels function as emergency release valves in vivo,
and the mechanosensitivity of pure MscL channel protein in vitro. a, An
E. coli cell in a normal environment (left) and in the rain (or upon dilution in
the laboratory, right). A bacterium (shown as a rod), having adjusted its
cytoplasm to the relatively high osmolarity of the surrounding milieu
(shown in dark red, the red dots being solutes, not water), is confronted with
a sudden dilution of its environment upon the onset of rain (light red). Entry
of water (not shown) through the lipid bilayer swells the bacterium (now
oval-shaped) and stretches open the MS channels to jettison solutes (red
puffs), enabling it to reach a new equilibrium and escaping osmolysis (and
returns to being rod-shaped). b, Purified MscL protein is reconstituted into
multilamellar liposomes after replacing the detergent with lipids. Induced
liposome blisters can be sampled with a patch-clamp pipette. A suction
applied to the pipette (broad open arrow) creates tension (small filled
arrows) in the membrane patch and activates MscL proteins. The increase in
the number of channel openings in a patch (shown as unitary-conductance
steps at the marked levels) is evident when the suction applied to such a
patch of lipid bilayer increases from 30 mm Hg to 40 mm Hg
(4  104
dyn cm2 2
to 5.3  104
dyn cm22
) (modified from ref. 15).
REVIEWS NATURE|Vol 436|4 August 2005
648
 2005 Nature Publishing Group
amphipaths to the membrane of a red blood cell causes it to form
cups, while adding anionic amphipaths causes it to form bulges (to
crenate)28
. Regardless of the cause, changes in the bilayer's geometry
will distort the force profile contained within it. Indeed, these agents
have been shown to reversibly activate MscL and MscS. The potency
of the amphipaths to activate such channels is proportional to their
lipid solubility29
, and they are effective only when added to one
monolayer but not both13
. The shapes of the added lipids are
important, as evident from the behaviour of gramicidin A upon
bilayer modifications uncovered by Andersen and co-workers30
. The
usual bilayer-forming phospholipids can be approximated as rods
(Fig. 4, shown in red), and the micelle-forming lysophospholipids
(with a single fatty-acid chain) as cones (Fig. 4, blue). Polyunsaturated
fatty acids (PUFAs), such as arachidonic acid (AA, a precursor
of prostaglandins), can be regarded as inverted cones with smaller
heads than tails (Fig. 4, green). Cones, or inverted cones, differentially
entered into one monolayer can change the local curvature and
the internal force profile, redistributing the tension between the two
leaflets (Fig. 3a). Indeed, the addition of lysophosphatidylcholine
triggers the opening of MscS (ref. 13). Structurally diverse anaesthetics,
which are all lipid soluble, have been theorized to change the
bilayer force profile31
. Indeed, experiments have shown that procaine
and tetracaine can activate MscS (ref. 29).
Eukaryotic MS channels
Plants (for example, Arabidopsis thaliana) have clear homologues of
MscS. Most animal cell membranes present MS conductances, but
only a few have been traced to known gene products. Although these
differ from MscS and MscL in sequence, some of their properties are
quite similar. Patel and co-workers found that the polymodal K
channel of mammals, TREK-1 (two-pore domain weak inwardrectifying
(TWIK)-related K
channel), can be activated by both
force and osmolarity. Similar to MscS and MscL, it is also activated by
the bulge-forming amphipathic chemicals such as trinitrophenol
(called crenaters), but inhibited by the cup-forming amphipaths
such as chlorpromazine32,33
. The cone-shaped lysophosphatidylcholine
activates it, and the exaggerated cone lysophosphatidylinositol
activates even more effectively32,34
. Structurally diverse anaesthetics
such as chloroform, halothane, isoflurane and diethyl-ether also
activate TREK-1 (ref. 35). Negatively charged lipids such as PIP2
(phosphatidylinositol 4,5-bisphosphate) or phosphatidic acid, when
presented to the membrane inner leaflet, also activate TREK-1
(ref. 36), and the charged cone lysophosphatidic acid strongly
activates it37
.
MS and some other types of channels are inhibited by the small
lanthanide Gd3
(refs 38, 39). The mechanisms of this inhibition are
complex and include its interaction with the lipid bilayer39,40
. Sachs
and others found an amphipathic 34 L-amino-acid peptide41
in
a tarantula venom to inhibit the MS conductance in cultured
mammalian cells42
. They also found the synthetic enantiomer of 34
D-amino-acids to be just as effective43
, and concluded that the
peptide could not have bound to a channel protein lock-and-keylike,
but entered the bilayer to affect the channel's surrounding
environment.
Eukaryotic cells have extensive cytoskeletons near the bilayer,
which are often assumed to be the transmitters of force. One needs
to examine this assumption carefully. Hamill et al. examined the MS
channels in the complex cell surface of Xenopus oocytes44
. From this
surface, they induced blebs that had little or no cytoskeletal elements
and continued to encounter MS-channel activities in the bleb
membranes16,44
, and it was from such membranes they now
traced the channel activities to TRPC1 (transient receptor-potential
canonical 1; ref. 45; discussed below). The cortical cytoskeleton
network often folds a large excess of membrane bilayer into microvilli
or cavaeolae. This network is far more extendable than the bilayer
itself, allowing cells to swell without increasing the total bilayer
area or tension, which explains why MS currents sometimes cannot
be found in whole cells but only in excised patches, where the
cytoskeleton is lost46,47
.
Animal sensory cells often have organized microtubules such as
the ciliary axoneme (see below). There are also specialized microtubular
processes along the length of the long touch-sensing cells in
Caenorhabditis elegans studied by Chalfie and co-workers48,49
. The
touch-insensitive (loss-of-function) or touch-cell degeneration
(gain-of-function)48­50
worm phenotypes led elegantly to the identification
of a series of mechanosensitivity genes (the mec genes)
Figure 2 | Opening MscL in E. coli. Helical segments (S1, S2, S3) and
transmembane helices (M1, M2) in one MscL subunit, as deduced from
sequence7
and other analyses9
(left). Side (upper centre) and top (lower
centre) views of the closed channel backbone structure of the E. coli MscL
protein, by analogy to the crystal structure of the M. tuberculosis MscL
homologue10
. The open structure deduced from both modelling and
experimentation11
(right). Unlike MthK, the prokaryotic K
channel that is
equipped with a second constriction (the K
filter), MscL is like the
acetylcholine receptor/channel, in which the open gate doubles as the filter.
Here the opening is huge (,30 A° in diameter): befitting its ability to release
solutes indiscriminately (as shown in Fig. 1a). The work to increase the area
under tension constitutes the free energy difference that partitions the open
and closed states. (Modified from ref. 11.)
NATURE|Vol 436|4 August 2005 REVIEWS
649
 2005 Nature Publishing Group
including mec-4 and mec-10. These genes encode channel subunits
akin to the epithelial Na
channels, and conduct channel currents in
frog oocytes51,52
. The channels, and their associated components,
form punctate clusters that are evenly spaced along the worm's
microtubular processes. Among the MECs are an extracellular matrix
protein (MEC-1) and special tubulins (MEC-7, MEC-12). Previous
models described a trans-cellular complex, similar to the original
trapdoor model for the vertebrate hair cell (now modified, see
below), in which the displacement of the matrix, resisted by the
microtubules, gates the channels in between48,49
. Recent analyses
showed that MEC-1, and at least two other gene products in the
matrix, are needed to form the punctate clusters. However, removal
of the special microtubules through mutation have only limited or no
effect on the structure or the function of these clusters53,54
. These
observations, and the fact that the transduction current is reduced
but not abolished in mec-7 mutants55
, have now questioned whether
the MS channel is directly gated by tethering to microtubules53
.
Tethering to rigid elements, even if true, does not necessarily imply
force transmission. Current cell biology teaches us that meaningful
protein­protein contacts, whether ephemeral or long-lasting, are the
norm. For example, the signalling machinery of the Drosophila
photoreceptor, not known for its mechanosensitivity properties, is
comprised of rhodopsin, G protein, enzymes and channels, all tied
together into a "transducisome"56
or "signalplex"57
tethered to an
actin cytoskeleton, which serves to deploy and station the complex
near the cell surface.
What trips TRP channels?
Several MS channels making recent news belong to the TRP super-
family45,58,59
, the founding member of which was identified in a
near-blind Drosophila with a transient receptor-potential in its
electroretinogram60
. Forward genetics starting from mechanoinsensitive
phenotypes led us to these TRP channels without preconceived
bias. This has happened six times recently--in the
worm61,62
, fly63,64
, mouse65
and human66
--and cannot be a coincidence.
At least seven homologues of the TRPs identified in this way
then became candidates in the ensuing investigations and have now
also been tied to mechanosensation.
TRPV4 (transient receptor-potential vanilloid-4; previously called
the vanilloid receptor-related osmotically activated channel, or VROAC),
the most studied mammalian MS TRP channel, is found in
many tissues including the circumventricular organs of the central
nervous system and inner-ear hair cells. The heterologously
expressed whole-cell current through TRPV4 can be activated by
cell inflation with pressure67
, by mechanical shear force from bath
flow68
, or by mild hypo-osmotic challenges69,70
. Deleting the three
amino-terminal ankyrin repeats does not significantly impair
TRPV4's response to hypo-osmolarity69
. The Bargmann laboratory
found the first MS TRPV through osm-9-mutant worms, which fail
to recoil from an osmolar solution or a nose touch71
. Worms with
normal OSM-9 channels, but deficient in the synthesis of a set of
20-carbon PUFAs (including AA, Fig. 4), showed similar deficits in
behaviour that could be restored by a dietary supplement of PUFAs.
One such PUFA acts as an irritant that appears to directly activate
the channel72
. The rat TRPV4 is only 24% identical to the worm's
OSM-9. Nonetheless, Liedtke et al. found that a trpv4 transgene
complements the defects of the osm-9-mutant worm, and that the
restored behaviours have a threshold and temperature optimum of a
warm-blooded animal73
. Furthermore, a version of TRPV4 deleted of
both its amino- and carboxy-terminal cytoplasmic domains, the
presumed cytoskeleton-binding sites, remained able to complement
the phenotype73
. The single-channel current of TRPV4 has received
far less scrutiny. Unitary conductances of 310 pS (ref. 69), 60 pS
(ref. 74), or 30 and 88 pS (ref. 70) have been variously reported from
similar experiments. The only report on attempts to activate TRPV4
unitary conductance by direct patch suction was negative70
. In
contrast, a 20 pS MS conductance (in frog Ringer's solution) from
oocyte membranes has recently been traced to TRPC1 through
liposome reconstitution by Maroto et al.45
, who also showed that
heterologously expressed human TRPC1 produces MS unitary currents.
Although reconstitution of purified TRPC1 has not yet been
reported, this work comes closest to showing that certain TRP
channels receive their gating force directly from the lipids.
Channels associated with mechanosensation appear in nearly all
TRP subfamilies: TRPV, TRPC, TRPA (ankyrin-like), TRPP (polycystin),
TRPN (NOMPC, or no mechanoreceptor potential C),
TRPY (yeast) and probably TRPML (mucolipins), each having rather
different cytoplasmic domains, suggesting that these domains are not
Figure 3 | The intrinsic forces in the lipid bilayer, and how applied forces
can open MS channels. a, The intrinsic force profile plotted as its direction
and magnitude along the depth of the bilayer (left)23
, and a cartoon of a
channel protein in section (right), showing how the sharp tension (narrow
arrows) near the lipid necks balanced by more diffused pressure nearby
(broad arrows) is exerted on the channel­lipid interface (red). b, The forces
at the crucial channel­lipid interface (red) will change when the bilayer
(green) is stretched or bent (left), or when the channel is displaced from the
bilayer through a tether like an elevator (right). It is also possible that the
tether, through ancillary proteins, pulls on the lipids surrounding the
channel (not shown). In all cases, changes in the force profile at the interface
(red) can become the ultimate trigger for the channel's conformational
change.
Figure 4 | The shape of bilayer components affects its geometry and
intrinsic forces. a, Bilayer-forming phospholipids (shown in red), such as
phosphatidylcholine (PC), can be approximated as rods. Micelle-forming
lysophospholipids (blue), such as lysophosphatidylcholine (LPC) with only
one fatty acid chain, can be regarded as cones. Polyunsaturated fatty acids
(green), such as arachidonic acid (AA), approximate inverted cones. b, The
differential addition of cone-shaped lipids (or other amphipaths, not shown)
into one of the monolayers can alter the shape, and therefore the intrinsic
forces, of the bilayer. (Modified from ref. 32.)
REVIEWS NATURE|Vol 436|4 August 2005
650
 2005 Nature Publishing Group
the crux of mechanosensitivity. TRP channels are polymodal.
Heterologously expressed TRPV4, for example, can also be activated
by heat, by a phorbol ester, by anandamide and by AA Fig. 4). The
well-known heat-sensing vanilloid receptor, TRPV1, identified in the
Julius laboratory75
, is also activated by low pH and endogenous
inflammatory ligands. The trpv1 knockout mice and their bladder
urothelium have abnormal response to hypo-osmolarity76
. Though a
universal switch for different mode of stimuli may be appealing, there
is evidence that different stimuli use different pathways to open
TRPV4 (ref. 77).
Strings attached--the roles of tethers
Several TRP channels have now been located within some of the
complex MS organs. For example, in the fly chordotonal organ, a
matrix material presses against the dendritic cap on top of the
sensory cilium, which swells into a dilation about one-third of
the way from the tip. Kim, Kernan, and their co-workers found
that the only two TRPV channels encoded within the Drosophila
genome, NAN (encoded by the Nanchung gene)78
and IAV (encoded
by the Inactive gene)79
, apparently form heteromeric channels that
transduce vibrations into receptor potentials. NAN and IAV proteins
are found on the cilia but not the rest of the neuron, and it is
restricted at and below the ciliary dilation, some distance from the tip
(Fig. 5a). How the movement of the cap-cell matrix translates into
the forces experienced by the IAV-NAN channel in the ciliary
membrane that lies below it is unclear. The NAN (ref. 78) and IAV
(ref. 79) proteins have been individually expressed in cultured cells,
where they confer hypo-osmotically induced whole-cell current and a
rise in cytoplasmic Ca2
levels. A parsimonious interpretation would
be that the channel experiences the vibration as a stretch along the
membrane plane. Whether there are other proteins that tie them to
the axoneme, whether such tethers transmit force, and whether there
are roles for other channel subunits (for example, NOMPA, or no
mechanoreceptor potential A, located at the distal end of the
cilium)79
await clarification.
The vertebrate hair cell is a clear case in which the gating force is
passed on to the MS channel through a tether; though it is debatable
whether the tether directly pulls a certain domain from the rest of the
channel protein, which is held in place by a resistive force, as
originally proposed. Molecular identifications recently enabled
great strides in this system. First, TRPN was found in sensory hair
cells, and to be required for hearing and balance in zebrafish80
. Then,
cadherin 23 was discovered to be a major component of the tip
link81,82
, and judged to be too stiff to comprise the gating spring.
Recently, Corey et al.83
showed that TRPA1 messenger RNA appears
at the appropriate time during hair-cell development, and that
TRPA1-expression knock-down curtails transduction in vivo.
TRPA1 protein is found at the upper part of the stereocilia, though
not just at the very tip. It is also located in the pericuticular zone
(perhaps for secretion) and, when present, the entire kinocilium (the
true cilium with a microtubular axoneme) (Fig. 5b). With such rapid
progress, we are looking forward to the identification of other gene
products in this system in the very near future. Meanwhile, fleshing
out the biophysical scheme with the molecules so far identified
has already led to modifications of the familiar `trapdoor' model of
hair-cell mechano-transduction (Fig. 5c, left). Calculation and simulation
show that the long N-terminal ankyrin repeats of TRPA1 are
compatible with the elastic property of the gating spring83,84
(Fig. 5c,
right). (The cartoons in Fig. 5c, and similar representations elsewhere,
should not be taken literally as the site(s) and the nature of
string attachment, in addition to the number and identity of channel
subunits and other elements, are unknown. The upper tether can be
attached to the gating springs beneath the membrane instead of the
channel body.) In the more thoroughly examined multimeric channels,
such as MthK (the K
channel of Methanobacterium thermoautotrophicum;
ref. 85), MscL (ref. 12) and MscS (ref. 14), the
channel gates open like the iris of a camera. It is not yet clear how
the one-dimensional movement of a stereocilium leads to the twodimensional
opening of the TRP tetramers on the side of the cilium,
and whether or how the ciliary membrane may move during
transduction. TRPA1 is also expressed in neurons of the dorsal
root ganglia, the trigeminal nerve, and photoreceptors, making one
wonder for which functions the ankyrin repeats serve in these other
locations. A role of ankyrin in assembling a TRP tetramer has been
suggested86
.
Returning to the main theme of this Review, the vertical movement
of a channel by the tip link does not necessarily negate the
involvement of the lipid bilayer. It seems possible that such `elevator'like
movement would still eventually displace the channel with
respect to the bilayer's intrinsic force profile (Fig. 3b, right). The
mismatch and asymmetry produced by the displacement can, in fact,
Figure 5 | TRP channels in auditory sensory cells. TRP channels have been
located in complex auditory sensory cells, even though the mechanism by
which ciliary vibrations (arrow pairs) lead to the iris-like opening of the
channels on the side of the cilia is not clear. a, The antennal chordotonal
organ of Drosophila. CM, cap-cell matrix; DC, dendritic cap; CD, ciliary
dilation. Red marks the location of NAN (a TRPV-type channel subunit
encoded by the Nanchung gene). (Redrawn from ref. 78.) b, Avertebrate hair
cell. St, stereocilia; K, kinocilium; PZ, pericuticular zone. Red marks the
location of TRPA1. (Modified from ref. 83.) c, Models of the vertebrate haircell
transduction channel. Molecular identifications have transformed the
biophysical trapdoor model (left) to one with a TRPA channel and a stiff
cadherin-containing tip link (right). The elastic element of transduction is
now assigned to the ankyrin repeats in the four (presumably) TRPA
subunits83,99
(shown as coils), which are presumed to be attached to
cytoskeleton and/or myosin (not shown). This current model is compatible
with one in which the displacement of the channel protein, with respect to
the lipid bilayer, ultimately triggers the channel conformation change as
shown in Fig. 3b, right. However, none of these models should be taken
literally since we do not yet know the true composition of the transduction
channel(s) and how the various channel components contact each other and
the lipids. See the main text for some possible variations.
NATURE|Vol 436|4 August 2005 REVIEWS
651
 2005 Nature Publishing Group
be the final mechano-energetic trigger for the required channelconfiguration
change. It is even possible that the tether pulls on the
rim of an elastic carrel, say, and passes the force through the lipids
that surround the channel (Fig. 3b, left), but this possibility has not
been investigated. The `trapdoor' (Fig. 5c, left) is usually interpreted
as the separation of individual protein domain(s) from the rest of the
channel protein by mechanical work17
. The `elevator' entails displacing
the entire channel protein from its normal lipid environments,
generating tension around the entire circumference, and leading to
an iris-like opening (Fig. 3b, right). Either the trapdoor or the
elevator model can accommodate the mechanical resistive and elastic
elements traditionally described.
It is difficult to imagine TRPA1 being indifferent to the lipids,
given its activation by lipid-like compounds. Gillespie and coworkers
showed that PIP2 localizes towards the tip of the hair
bundle and is required for both the mechanical transduction current
and its adaptations87
. The mouse TRPA1 was first reported to be a
sensor of noxious cold88
. It is also activated by bradykinin, and the
oils of mustard, cinnamon, wintergreen and the like, as well as
cannabinoids89,90
.
Solute senses versus solvent senses
Specialized sensory cilia develop from embryonic primary cilia, and
have evolved from motile cilia similar to the ones still found in
Paramecium, Chlamydomonas and so on. Can we trace the origin of
TRP-based mechanosensation beyond motile protists? An MS TRP
channel is found in the vacuolar membrane of yeast91
, where it
detects osmotic forces in vitro92
and in vivo93,94
. Because all cells have
to deal with osmotic force, it may hold a key to the evolutionary
origin(s) of mechanosensation.
MscL and MscS proteins are found in most free-living Bacteria and
Archaea. Thus, the principle of mechanical gating by forces from the
lipid bilayer most likely evolved before the divergence of these two
domains of life some 3.5 billion years ago. It makes teleological sense
for these devices to have evolved early on. When early cells separate
two solutions, and horde solutes into one, a water gradient--and
therefore a turgor--must develop on the chemiosmotic partition.
This turgor has apparently been used ever since, as all extant cells
today still have to be turgid to be in a growing steady state. This
turgidity helps to break the bonds in the network of rigid elements
(membrane, cell wall, extracellular matrix, cytoskeleton and so forth)
so that new material can be inserted. Off steady states--the sudden
large rise in turgor at the onset of rain (over-hydration) and the large
fall in turgor in prolonged hot sun (dehydration)--are likely to have
exerted selective pressure on early cells to evolve mechanosensors
such as the ancestors of TRP, MscS and MscL. Once the basic
principle of activation by lipid forces is employed, it seems unlikely
that nature would abandon the principle in the detection of other
forces later on in evolution, even as newer generations of protein
types out-compete the old. As reviewed above, many extant TRP
channels in animals still seem to respond to osmotic forces exerted on
the lipid bilayer even though their relatives are specialized for
hearing, balance, touch, or texture.
Today, we find a myriad of surface receptors for irritants, odorants,
hormones, neurotransmitters and growth factors on cell surfaces.
They are not sequence homologues, but they could have originated
from one or a few receptors by divergent, as well as convergent,
evolution. More importantly, the same physicochemical principle of
lock-and-key fitting underlies all ligand sensing. Yet one cannot
imagine a protein with a lock-and-key water-binding site that
discriminates the milli-osmolar differences that cells do, while
water exists in tens of molar. Rapid changes in water concentration
must therefore be sensed with a different mechanism. From this
perspective, it is not difficult to imagine that a parallel myriad of
mechanosensors may have evolved from some simple osmotic-forcesensing
device in early membranes95
. Figure 6a shows an `ur-cell' with
the two types of receptors for the various solutes (for example,
nutrients and wastes) and for the one solvent (water). Figure 6b
shows a schematic representation of how different senses might have
evolved throughout the 3.5 billion years of evolution. This is not a
molecular dendrogram. The branching pattern has no precise meaning
except that the senses were few in number in the beginning, and
that sensing of nutrients and water is fundamental, ancient and
disparate. This diagram emphasizes the distinctness of the two classes
of sensations that originated to deal with solutes versus solvent--the
two basic ingredients of life's chemistry.
Complexities and unknowns
The disparity between solute sensing and solvent sensing can help
clarify our thinking. Many solute (ligand) receptors do not require a
membrane. It is also `crystal clear' (from X-ray diffraction studies)
that some MS channels have no specific ligand-binding sites (for
example, MscL, MscS). These examples should not lead us to assert
that no receptors employ both principles. Given nature's propensity
to tinker opportunistically, it is even likely for some receptor
proteins to evolve specific ligand-binding pockets that face the two
aqueous compartments, as well as a lipid-facing surface that transduces
forces. This complexity may explain certain polymodality74
or
Figure 6 | The disparate sensing of solutes and solvent. a, A diagram of an
imaginary early cell equipped with two types of receptors that are required to
sense solutes and solvents--the two ingredients of life's chemistry. The dots
in the grey background represent water molecules (the solvent) and the red
circles represent solutes (molecules dissolved in water). When a cell
accumulates solutes, the internal water concentration is reduced and the
tendency of water to enter the cell results in a turgor. Both the lock-and-key
type of receptors (red) for different solutes (ligands), as well as the turgor
sensors (blue) for water (the solvent), are needed for even an early cell to
survive. b, A hypothetical diagram (not to be mistaken for phylogenetic
trees) on the grouping of various senses that emphasizes the discrete
separations of the lock-and-key type of sensing of the solutes (red) from the
force-from-bilayer type of sensing of the solvent (blue). A further
description can be found in the text. (Modified from ref. 100.)
REVIEWS NATURE|Vol 436|4 August 2005
652
 2005 Nature Publishing Group
other differences (for example, the TRPV-dependent behaviour of
live worms requires certain features of PUFAs but not necessarily
AA72
, while heterologously expressed TRPV4 in vitro seems to require
an AA metabolite74
). Complexities can also arise from possible
`signalling lipids' that may act at the protein­lipid­water junction.
For example, adding negative charges to the inner leaflet with PIP2
(ref. 87) will have local electrostatic effects36,37
as well as secondary
effects on lipid distribution and bilayer force profile.
In On the Soul (ref. 96), Aristotle argued that there could be no
more than five senses, and folded the senses of hot and cold into
touch. He might have been on to something. The detection of heat
and that of impact seem to convolve in biology. Heat-induced bilayer
rearrangement may alter the membrane tension that gates the
polymodal TRP channels97
. Teleologically, have they been designed
to sum heat and force (for example, into `pain')? Mechanistically, can
we separate the force sensor from the heat sensor within the same
protein? The truism that everything is mechanical, and heatsensitive,
is not helpful. How can a protein­lipid­water complex be
designed to be especially sensitive to a rise in thermal agitation? And
how can a similar complex become sensitive to its fall? How can
some TRPs be constructed to have an unusually high Q10 value (in
excess of 10)? If the thermosensor is coupled to a voltage sensor98
,
what is the coupling mechanism? Doesn't the temperature sensitivity
of a channel­lipid complex depend on the entropy of its
hydrophobic interactions with water? Why are the temperature
mimics (such as capsaicin, menthol and ginger) oily? Why are the
agents that numb the sensation amphipathic? Why is the potency
of different general anaesthetics proportional to their solubility in
olive oil (the Meyer­Overton rule)? Could the protein­lipid­water
junction hold even more secrets? Much work lies ahead if the
answers to these questions are to be found; perhaps, just like after
a long spell of rain, the floodgates of knowledge are about to be
opened.
1. Britten, R. J. & McClure, F. T. The amino acid pool in Escherichia coli. Bacteriol.
Rev. 26, 292­-335 (1962).
2. Berrier, C., Coulombe, A., Szabo, I., Zoratti, M. & Ghazi, A. Gadolinium ion
inhibits loss of metabolites induced by osmotic shock and large stretchactivated
channels in bacteria. Eur. J. Biochem. 206, 559­-565 (1992).
3. Martinac, B., Buechner, M., Delcour, A. H., Adler, J. & Kung, C. Pressuresensitive
ion channel in Escherichia coli. Proc. Natl Acad. Sci. USA 84,
2297­-2301 (1987).
4. Levina, N. et al. Protection of Escherichia coli cells against extreme turgor by
activation of MscS and MscL mechanosensitive channels: identification of
genes required for MscS activity. EMBO J. 18, 1730­-1737 (1999).
5. Martinac, B., Delcour, A. H., Buechner, M., Adler, J. & Kung, C. Advances in
Comparative and Environmental Physiology 3­-18 (Springer, Heidelberg, 1992).
6. Delcour, A. H., Martinac, B., Adler, J. & Kung, C. Modified reconstitution
method used in patch-clamp studies of Escherichia coli ion channels. Biophys. J.
56, 631­-636 (1989).
7. Sukharev, S. I., Blount, P., Martinac, B., Blattner, F. R. & Kung, C. A largeconductance
mechanosensitive channel in E. coli encoded by mscL alone.
Nature 368, 265­-268 (1994).
8. Sukharev, S. I., Martinac, B., Blount, P. & Kung, C. Functional reconstitution as
an assay for biochemical isolation of channel proteins: application to the
molecular identification of a bacterial mechanosensitive channel. Methods: A
Companion to Methods in Enzymology 6, 51­-59 (1994).
9. Blount, P. et al. Membrane topology and multimeric structure of a
mechanosensitive channel protein of Escherichia coli. EMBO J. 15, 4798­-4805
(1996).
10. Chang, G., Spencer, R. H., Lee, A. T., Barclay, M. T. & Rees, D. C. Structure of
the MscL homolog from Mycobacterium tuberculosis: a gated mechanosensitive
ion channel. Science 282, 2220­-2226 (1998).
11. Sukharev, S., Betanzos, M., Chiang, C. S. & Guy, H. R. The gating mechanism of
the large mechanosensitive channel MscL. Nature 409, 720­-724 (2001).
12. Perozo, E., Cortes, D. M., Sompornpisut, P., Kloda, A. & Martinac, B. Open
channel structure of MscL and the gating mechanism of mechanosensitive
channels. Nature 418, 942­-948 (2002).
13. Perozo, E., Kloda, A., Cortes, D. M. & Martinac, B. Physical principles
underlying the transduction of bilayer deformation forces during
mechanosensitive channel gating. Nature Struct. Biol. 9, 696­-703 (2002).
14. Bass, R. B., Strop, P., Barclay, M. & Rees, D. C. Crystal structure of Escherichia
coli MscS, a voltage-modulated and mechanosensitive channel. Science 298,
1582­-1587 (2002).
15. Sukharev, S. I., Blount, P., Martinac, B. & Kung, C. Mechanosensitive channels
of Escherichia coli: the MscL gene, protein, and activities. Annu. Rev. Physiol. 59,
633­-657 (1997).
16. Hamill, O. P. & Martinac, B. Molecular basis of mechanotransduction in living
cells. Physiol. Rev. 81, 685­-740 (2001).
17. Sukharev, S. & Corey, D. P. Mechanosensitive channels: multiplicity of families
and gating paradigms. Sci. STKE doi:10.1126/stke.2332004eg7 (2004).
18. Blount, P. Molecular mechanisms of mechanosensation: big lessons from small
cells. Neuron 37, 731­-734 (2003).
19. Sukharev, S. I., Sigurdson, W. J., Kung, C. & Sachs, F. Energetic and spatial
parameters for gating of the bacterial large conductance mechanosensitive
channel, MscL. J. Gen. Physiol. 113, 525­-540 (1999).
20. Sukharev, S. I., Martinac, B., Arshavsky, V. Y. & Kung, C. Two types of
mechanosensitive channels in the Escherichia coli cell envelope: solubilization
and functional reconstitution. Biophys. J. 65, 177­-183 (1993).
21. Ou, X. R., Blount, P., Hoffman, R. J. & Kung, C. One face of a transmembrane
helix is crucial in mechanosensitive channel gating. Proc. Natl Acad. Sci. USA 95,
11471­-11475 (1998).
22. Maurer, J. A. & Dougherty, D. A. Generation and evaluation of a large
mutational library from the Escherichia coli mechanosensitive channel of large
conductance, MscL: implications for channel gating and evolutionary design.
J. Biol. Chem. 278, 21076­-21082 (2003).
23. Cantor, R. S. Lateral pressures in cell membranes: a mechanism for modulation
of protein function. J. Phys. Chem. 101, 1723­-1725 (1997).
24. Cantor, R. S. The influence of membrane lateral pressures on simple geometric
models of protein conformational equilibria. Chem. Phys. Lipids 101, 45­-56
(1999).
25. Lindahl, E. & Edholm, O. Spatial and energetic-entropic decomposition of
surface tension in lipid bilayers from molecular dynamics simulations. J. Chem.
Phys. 113, 3882­-3893 (2000).
26. Gullingsrud, J. & Schulten, K. Gating of MscL studied by steered molecular
dynamics. Biophys. J. 85, 2087­-2099 (2003).
27. Wiggins, P. & Phillips, R. Analytic models for mechanotransduction: gating a
mechanosensitive channel. Proc. Natl Acad. Sci. USA 101, 4071­-4076 (2004).
28. Sheetz, M. P. & Singer, S. J. Biological membranes as bilayer couples. A
molecular mechanism of drug-erythrocyte interactions. Proc. Natl Acad. Sci.
USA 71, 4457­-4461 (1974).
29. Martinac, B., Adler, J. & Kung, C. Mechanosensitive ion channels of E. coli
activated by amphipaths. Nature 348, 261­-263 (1990).
30. Lundbaek, J. A., Maer, A. M. & Andersen, O. S. Lipid bilayer electrostatic
energy, curvature stress, and assembly of gramicidin channels. Biochemistry 36,
5695­-5701 (1997).
31. Cantor, R. S. The lateral pressure profile in membranes: a physical mechanism
of general anesthesia. Toxicol. Lett. 100­-101, 451­-458 (1998).
32. Patel, A. J., Lazdunski, M. & Honore, E. Lipid and mechano-gated 2P domain
K
channels. Curr. Opin. Cell Biol. 13, 422­-428 (2001).
33. Patel, A. J. et al. A mammalian two pore domain mechano-gated S-like K
channel. EMBO J. 17, 4283­-4290 (1998).
34. Maingret, F., Patel, A. J., Lesage, F., Lazdunski, M. & Honore, E.
Lysophospholipids open the two-pore domain mechano-gated K
channels
TREK-1 and TRAAK. J. Biol. Chem. 275, 10128­-10133 (2000).
35. Patel, A. J. & Honore, E. Anesthetic-sensitive 2P domain K
channels.
Anesthesiology 95, 1013­-1021 (2001).
36. Chemin, J. et al. A phospholipid sensor controls mechanogating of the K
channel TREK-1. EMBO J. 24, 44­-53 (2005).
37. Chemin, J. et al. Lysophosphatidic acid-operated K
channels. J. Biol. Chem.
280, 4415­-4421 (2005).
38. Yang, X. C. & Sachs, F. Block of stretch-activated ion channels in Xenopus
oocytes by gadolinium and calcium ions. Science 243, 1068­-1071 (1989).
39. Hamill, O. P. & McBride, D. W. Jr The pharmacology of mechanogated
membrane ion channels. Pharmacol. Rev. 48, 231­-252 (1996).
40. Ermakov, Y. A., Averbakh, A. Z., Yusipovich, A. I. & Sukharev, S. Dipole
potentials indicate restructuring of the membrane interface induced by
gadolinium and beryllium ions. Biophys. J. 80, 1851­-1862 (2001).
41. Suchyna, T. M. et al. Identification of a peptide toxin from Grammostola
spatulata spider venom that blocks cation-selective stretch-activated channels.
J. Gen. Physiol. 115, 583­-598 (2000); erratum J. Gen. Physiol. 117, 371 (2001).
42. Guharay, G. & Sachs, F. Stretch-activated single ion channel currents in tissuecultured
embryonic chick skeletal muscle. J. Physiol. (Lond.) 352, 685­-701
(1984).
43. Suchyna, T. M. et al. Bilayer-dependent inhibition of mechanosensitive
channels by neuroactive peptide enantiomers. Nature 430, 235­-240 (2004).
44. Zhang, Y., Gao, F., Popov, V. L., Wen, J. W. & Hamill, O. P. Mechanically gated
channel activity in cytoskeleton-deficient plasma membrane blebs and vesicles
from Xenopus oocytes. J. Physiol. (Lond.) 523, 117­-130 (2000).
45. Maroto, R. et al. The role of TRPC1 in forming the mechanosensitive cation
channel in frog oocytes. Nature Cell Biol. 7, 179­-185 (2005).
46. Morris, C. E. & Horn, R. Failure to elicit neuronal macroscopic
mechanosensitive currents anticipated by single-channel studies. Science 251,
1246­-1249 (1991).
47. Zhang, Y. & Hamill, O. P. On the discrepancy between whole-cell and
membrane patch mechanosensitivity in Xenopus oocytes. J. Physiol. (Lond.) 523,
101­-115 (2000).
NATURE|Vol 436|4 August 2005 REVIEWS
653
 2005 Nature Publishing Group
48. Ernstrom, G. G. & Chalfie, M. Genetics of sensory mechanotransduction. Annu.
Rev. Genet. 36, 411­-453 (2002).
49. Goodman, M. B. & Schwarz, E. M. Transducing touch in Caenorhabditis elegans.
Annu. Rev. Physiol. 65, 429­-452 (2003).
50. Bianchi, L. et al. The neurotoxic MEC-4(d) DEG/ENaC sodium channel
conducts calcium: implications for nevrosis initiation. Nature Neurosci. 7,
1337­-1344 (2004).
51. Goodman, M. B. et al. MEC-2 regulates C. elegans DEG/ENaC channels needed
for mechanosensation. Nature 415, 1039­-1042 (2002).
52. Chelur, D. S. et al. The mechanosensory protein MEC-6 is a subunit of the C.
elegans touch-cell degenerin channel. Nature 420, 669­-673 (2002).
53. Emtage, L., Gu, G., Hartwieg, E. & Chalfie, M. Extracellular proteins organize
the mechanosensory channel complex in C. elegans touch receptor neurons.
Neuron 44, 795­-807 (2004).
54. Zhang, S. et al. MEC-2 is recruited to the putative mechanosensory complex in
C. elegans touch receptor neurons through its stomatic-like domain. Curr. Biol.
14, 1888­-01896 (2004).
55. O'Hagan, R., Chalfie, M. & Goodman, M. B. The MEC-4 DEG/ENaC channel of
Caenorhabditis elegans touch receptor neurons transduces mechanical signals.
Nature Neurosci. 8, 43­-50 (2005).
56. Tsunoda, S. et al. A multivalent PDZ-domain protein assembles signalling
complexes in a G-protein-coupled cascade. Nature 388, 243­-249 (1997).
57. Montell, C. TRP trapped in fly signalling web. Curr. Opin. Neurobiol. 8, 389­-397
(1998).
58. Montell, C., Birnbaumer, L. & Flockerzi, V. The TRP channels, a remarkably
functional family. Cell 108, 595­-598 (2002).
59. Corey, D. P. New TRP channels in hearing and mechanosensation. Neuron 39,
585­-588 (2003).
60. Minke, B., Wu, C.-F. & Pak, W. L. Induction of photoreceptor voltage noise in
the dark in Drosophila mutant. Nature 258, 84­-87 (1975).
61. Colbert, H. A., Smith, T. L. & Bargmann, C. I. OSM-9, a novel protein with
structural similarity to channels, is required for olfaction, mechanosensation,
and olfactory adaptation in Caenorhabditis elegans. J. Neurosci. 17, 8259­-8269
(1997).
62. Barr, M. M. & Sternberg, P. W. A polycystic kidney-disease gene homologue
required for male mating behaviour in C. elegans. Nature 401, 386­-389 (1999).
63. Walker, R. G., Willingham, A. T. & Zuker, C. S. A Drosophila mechanosensory
transduction channel. Science 287, 2229­-2234 (2000).
64. Tracey, W. D. Jr, Wilson, R. I., Laurent, G. & Benzer, S. painless, a Drosophila
gene essential for nociception. Cell 113, 261­-273 (2003).
65. Di Palma, F. et al. Mutations in Mcoln3 associated with deafness and
pigmentation defects in varitint-waddler (Va) mice. Proc. Natl Acad. Sci. USA
99, 14994­-14999 (2002).
66. Mochizuki, T. et al. PKD2, a gene for polycystic kidney disease that encodes an
integral membrane protein. Science 272, 1339­-1342 (1996).
67. Suzuki, M., Mizuno, A., Kodaira, K. & Imai, M. Impaired pressure sensation in
mice lacking TRPV4. J. Biol. Chem. 278, 22664­-22668 (2003).
68. Gao, X., Wu, L. & O'Neil, R. G. Temperature-modulated diversity of TRPV4
channel gating: activation by physical stresses and phorbol ester derivatives
through protein kinase C-dependent and -independent pathways. J. Biol. Chem.
278, 27129­-27137 (2003).
69. Liedtke, W. et al. Vanilloid receptor-related osmotically activated channel
(VR-OAC), a candidate vertebrate osmoreceptor. Cell 103, 525­-535 (2000).
70. Strotmann, R., Harteneck, C., Nunnenmacher, K., Schultz, G. & Plant, T. D.
OTRPC4, a nonselective cation channel that confers sensitivity to extracellular
osmolarity. Nature Cell Biol. 2, 695­-702 (2000).
71. Tobin, D. et al. Combinatorial expression of TRPV channel proteins defines
their sensory functions and subcellular localization in C. elegans neurons.
Neuron 35, 307­-318 (2002).
72. Kahn-Kirby, A. et al. Specific polyunsaturated fatty acids drive
TRPV-dependent sensory signalling in vivo. Cell 119, 889­-900 (2004).
73. Liedtke, W., Tobin, D. M., Bargmann, C. I. & Friedman, J. M. Mammalian
TRPV4 (VR-OAC) directs behavioural responses to osmotic and mechanical
stimuli in Caenorhabditis elegans. Proc. Natl Acad. Sci. USA 100 (suppl. 2),
14531­-14536 (2003).
74. Watanabe, H. et al. Anandamide and arachidonic acid use epoxyeicosatrienoic
acids to activate TRPV4 channels. Nature 424, 434­-438 (2003).
75. Caterina, M. J. et al. The capsaicin receptor: a heat-activated ion channel in the
pain pathway. Nature 389, 816­-824 (1997).
76. Birder, L. A. et al. Altered urinary bladder function in mice lacking the vanilloid
receptor TRPV1. Nature Neurosci. 5, 856­-860 (2002).
77. Vriens, J. et al. Cell swelling, heat, and chemical agonists use distinct pathways
for the activation of the cation channel TRPV4. Proc. Natl Acad. Sci. USA 101,
396­-401 (2004).
78. Kim, J. et al. A TRPV family ion channel required for hearing in Drosophila.
Nature 424, 81­-84 (2003).
79. Gong, S. et al. Two interdependent TRPV chanel subunits, Inactive and
Nanchung, mediate hearing in Drosophila. J. Neurosci. 24, 9059­-9066 (2004).
80. Sidi, S., Friedrich, R. W. & Nicolson, T. NompC TRP channel required for
vertebrate sensory hair cell mechanotransduction. Science 301, 96­-99 (2003).
81. Siemens, J. et al. Cadherin 23 is a component of the tip link in hair-cell
stereocilia. Nature 428, 950­-955 (2004).
82. Sollner, C. et al. Mutations in cadherin 23 affect tip links in zebrafish sensory
hair cells. Nature 428, 955­-959 (2004).
83. Corey, D. P. et al. TRPA1 is a candidate for the mechanosensitive transductin
channel of vertebrate hair cells. Nature 432, 723­-730 (2004).
84. Howard, J. & Bechstedt, S. Hypothesis: a helix of ankyrin repeats of the
NOMPC-TRP ion channel is the gating spring of mechanoreceptors. Curr. Biol.
14, R224­-R226 (2004).
85. Jiang, Y. et al. The open pore conformation of potassium channels. Nature 417,
523­-526 (2002).
86. Erler, I., Hirnet, D., Wissenbach, U., Flockerzi, V. & Niemeyer, B. A. Ca2
selective
transient receptor potential V channel architecture and function
require a specific ankyrin repeat. J. Biol. Chem. 279, 34456­-34463 (2004).
87. Hirono, M., Denis, C. S., Richardson, G. P. & Gillespie, P. G. Hair cells require
phosphatidylinositol 4,5-bisphosphate for mechanical transduction and
adaptation. Neuron 44, 309­-320 (2004).
88. Story, G. M. et al. ANKTM1, a TRP-like channel expressed in nociceptive
neurons, is activated by cold temperatures. Cell 112, 819­-829 (2003).
89. Jordt, S. E. et al. Mustard oils and cannabinoids excite sensory nerve fibres
through the TRP channel ANKTM1. Nature 427, 260­-265 (2004).
90. Bandell, M. et al. Noxious cold ion channel TRPA1 is activated by pungent
compounds and bradykinin. Neuron 41, 849­-857 (2004).
91. Palmer, C. P. et al. A TRP homolog in Saccharomyces cerevisiae forms an
intracellular Ca2
-permeable channel in the yeast vacuolar membrane. Proc.
Natl Acad. Sci. USA 98, 7801­-7805 (2001).
92. Zhou, X. L. et al. The transient receptor potential channel on the yeast vacuole
is mechanosensitive. Proc. Natl Acad. Sci. USA 100, 7105­-7110 (2003).
93. Denis, V. & Cyert, M. S. Internal Ca(2  ) release in yeast is triggered by
hypertonic shock and mediated by a TRP channel homologue. J. Cell Biol. 156,
29­-34 (2002).
94. Zhou, X.-L., Loukin, S. H., Coria, R., Kung, C. & Saimi, Y. Heterologously
expressed fungal transient receptor potential channels retain
mechanosensitivity in vitro and osmotic response in vivo. Eur. Biophys. J. 34,
413­-422 (2005).
95. Kung, C., Saimi, Y. & Martinac, B. Current Topics in Membranes and Transport
145­-153 (Academic, New York, 1990).
96. Apostle, H. G. Aristotle's On The Soul (De Anima) (Translation) 42­-43
(Peripatetic, Crinnell, Iowa, 1981).
97. Clapham, D. E. TRP channels as cellular sensors. Nature 426, 517­-524 (2003).
98. Voets, T. et al. The principle of temperature-dependent gating in cold- and
heat-sensitive TRP channels. Nature 430, 748­-754 (2004).
99. Corey, D. P. & Sotomayor, M. Hearing: tightrope act. Nature 428, 901­-903
(2004).
100. Kung, C. in Evolution of the First Nervous Systems (ed. Anderson, P. A. V.)
203­-214 (Plenum, New York, 1990).
Acknowledgements I thank A. Anishkin, M. Chalfie, R. Fettiplace, W. J. Haynes,
S. Loukin, B. Martinac, Y. Saimi, A. O. W. Stretton and X.-L. Zhou for discussions
and criticisms. My laboratory is supported by the Vilas Trust of the University of
Wisconsin and by the NIH.
Author Information Reprints and permissions information is available at
npg.nature.com/reprintsandpermissions. The author declares no competing
financial interests. Correspondence and requests for materials should be
addressed to C.K. (ckung@wisc.edu).
REVIEWS NATURE|Vol 436|4 August 2005
654
 2005 Nature Publishing Group