Swelling-Activated Ca2+ Channels Trigger Ca2+ Signals in Merkel Cells

Merkel cell-neurite complexes are highly sensitive touch receptors comprising epidermal Merkel cells and sensory afferents. Based on morphological and molecular studies, Merkel cells are proposed to be mechanosensory cells that signal afferents via neurotransmission; however, functional studies testing this hypothesis in intact skin have produced conflicting results. To test this model in a simplified system, we asked whether purified Merkel cells are directly activated by mechanical stimulation. Cell shape was manipulated with anisotonic solution changes and responses were monitored by Ca2+ imaging with fura-2. We found that hypotonic-induced cell swelling, but not hypertonic solutions, triggered cytoplasmic Ca2+ transients. Several lines of evidence indicate that these signals arise from swelling-activated Ca2+-permeable ion channels. First, transients were reversibly abolished by chelating extracellular Ca2+, demonstrating a requirement for Ca2+ influx across the plasma membrane. Second, Ca2+ transients were initially observed near the plasma membrane in cytoplasmic processes. Third, voltage-activated Ca2+ channel (VACC) antagonists reduced transients by half, suggesting that swelling-activated channels depolarize plasma membranes to activate VACCs. Finally, emptying internal Ca2+ stores attenuated transients by 80%, suggesting Ca2+ release from stores augments swelling-activated Ca2+ signals. To identify candidate mechanotransduction channels, we used RT-PCR to amplify ion-channel transcripts whose pharmacological profiles matched those of hypotonic-evoked Ca2+ signals in Merkel cells. We found 11 amplicons, including PKD1, PKD2, and TRPC1, channels previously implicated in mechanotransduction in other cells. Collectively, these results directly demonstrate that Merkel cells are activated by hypotonic-evoked swelling, identify cellular signaling mechanisms that mediate these responses, and support the hypothesis that Merkel cells contribute to touch reception in the Merkel cell-neurite complex.


Introduction
The mechanical senses of touch, proprioception, hearing and balance are initiated by cells that transduce force into electrical signals. Though the mechanotransduction events underlying audition and balance have been extensively investigated [1,2], little is known about the signals that initiate the somatic mechanical senses of proprioception and touch [3,4]. Progress has been hampered by the anatomy of the somatosensory system. Mechanotransduction occurs in peripheral afferent terminals, which are heterogeneous and widely distributed throughout the body, making most somatosensory mechanoreceptors difficult to study directly.
Merkel cell-neurite complexes, which are cutaneous mechanoreceptors critical for fine shape and texture discrimination [4], have several characteristics that overcome these experimental challenges. Unlike most cutaneous touch receptors, the response patterns of these touch receptors have been identified in semiintact recording preparations: when the skin is displaced they generate the slowly adapting type I (SAI) response [5,6]. Furthermore, in transgenic Math1/nGFP mice, Merkel cells are specifically labeled with green fluorescent protein (GFP) [7], permitting enrichment of Merkel cells from dissociated skin using fluorescence-activated cell sorting (FACS) [8].
Since their discovery, epidermal Merkel cells have been proposed to be mechanosensory cells that transduce mechanical stimuli and then transmit sensory information to underlying sensory afferents [9]. This hypothesis stems from their location adjacent to sensory afferent terminals in highly touch sensitive areas of skin. This anatomy resembles that of hair cells, the forcesensitive epithelial cells of the inner ear that make synaptic contacts with afferent terminals. Moreover, Merkel cells express presynaptic molecules essential for synaptic transmission [8,10,11]. Finally, Merkel cells have functional VACCs [8,12], which trigger vesicle release at neuronal synapses. Despite this evidence, physiological experiments have failed to conclusively determine if Merkel cells are required for the SAI response [13]. Eliminating Merkel cells by laser ablation or genetic deficiency abolished the SAI response in some studies [14,15] but not others [16,17]. Previous reports exploring Merkel-cell mechanosensitivity in vivo and in vitro have been likewise inconclusive [18,19].
To determine if Merkel cells are mechanosensory cells, we asked if purified Merkel cells directly respond to mechanical stimuli in vitro. When the skin is displaced, forces must alter Merkel-cell shape, though the exact nature of this deformation is not known. To induce such a shape change in vitro, we chose to stimulate Merkel cells with osmotic stimuli because they permit simultaneous stimulation of many cells. Also, this is a robust stimulus that activates force-sensitive ion channels [20][21][22] as well as mechanosensitive cells such as hair cells and somatosensory neurons [23,24]. Moreover, MscL, the mechanosensitive channel whose structure and function is best understood, is directly activated by hypotonic stimuli in native cells [25]. Our findings demonstrate that hypotonic stimuli cause Ca 2+ influx in purified Merkel cells, and indicate that this Ca 2+ influx is initiated by swelling-activated ion channels. We used RT-PCR and pharmacology to identify candidate ion channels that may mediate this response. Our results demonstrate that Merkel cells are directly activated by a mechanical stimulus, which supports the hypothesis that they function as mechanoreceptors in Merkel cell-neurite complexes.

Cell preparation
All animal research was conducted according to protocols approved by Institutional Animal Care and Use Committee (IACUC) of Baylor College of Medicine (BCM) and the University of California, San Francisco (UCSF). Merkel cells were dissociated from the skin of postnatal day 3-6 (P3-P6) Math1/nGFP [7,8] mice after euthanasia by decapitation with sharp scissors. The skin from the body and face was dissected and washed in 10% Hibiclens (Regent Medical) and Hank's balanced salt solution (HBSS) supplemented with penicillin, streptomycin and amphotericin B. Tissue was cut into 1-cm 2 pieces and incubated for 1 h at 23uC in dispase (BD Biosciences) suspended to 25 U/mL in Ca 2+ and Mg 2+ free HBSS. The epidermis was peeled from the dermis with sharp forceps and incubated in 0.1% trypsin and 1 mM EDTA-4Na solution (Gibco) for 15 min with periodic vortexing. Trypsin was neutralized with fetal bovine serum (FBS) and cells were triturated with a 5-ml serological pipette. Cells were filtered with 70-and 40-mm cell strainers, spun at 4006g for 12-15 min and then resuspended in keratinocyte media (CNT-02, Chemicon) with 10% FBS. GFP-positive Merkel cells were enriched to approximately 85% by FACS into a landing media containing 50% FBS and 50% keratinocyte media (CNT-02, Chemicon). Merkel cells were spotted onto either collagen-coated coverslips for Ca 2+ imaging or collagen-coated eight-well chamber slides (LAB-TEK) for cell-volume analysis and grown with 5% CO 2 at 37uC in antibiotic-free keratinocyte media (CNT-02, Chemicon).

Live-cell Ca 2+ imaging
After 2 days in culture, Merkel cells were loaded for 20 min with 2 mM fura-2 acetoxymethyl ester (Molecular Probes) and 0.02% pluronic F-127 (Molecular Probes) in a modified Ringer's solution containing (in mM): 110 NaCl, 5 KCl, 10 HEPES (pH 7.4), 10 D-Glucose, 2 MgCl 2 , 2 CaCl 2 and 30 mannitol (290 mmol?kg 21 ). Cells were allowed to digest the ester bonds for 30 min and were imaged in modified Ringer's solution. Twenty percent hypotonic solutions contained all the same elements as modified Ringer's solution except mannitol. At these concentrations, mannitol did not fluoresce significantly in the fura-2 fluorescence spectra. To make 30% hypertonic solution, modified Ringer's solution was supplemented with an additional 45  Cells were viewed with a BX61WI epiflourescence upright microscope (Olympus) equipped with XLUMPlanFl 206, 0.95 NA and 606, 0.9 NA dipping objective lenses. Cells were illuminated with a 300-W Xenon lamp equipped with a highspeed excitation filter wheel (Sutter). Emission was captured with a cooled CCD camera (Hamamatsu). Data were acquired with Metafluor software of Meta Imaging series (version 6.4.7, Molecular Devices), and analyzed with custom algorithms written in Igor Pro (Version 5.03, Wavemetrics).
The Ca 2+ dissociation constant of fura-2 in Merkel cells was determined by performing a three point calibration in situ with solutions containing (in mM): 135 KCl, 2 MgCl 2 , 10 HEPES, and one of the following: 10 EGTA (for R min ), 10 CaCl 2 (for R max ), and 8.5 EGTA plus 1.5 CaCl 2 (for R mid , effective free [Ca 2+ ] = 0.9 mM as measured in vitro by fura-2 imaging). Merkel cells were rendered Ca 2+ -permeable by ionomycin (1 mM) and triton X-100 (0.01-0.015%). Cellular respiration was inhibited with 2 mM 2-deoxy-D-glucose to block active pumps. Only cells with stable F 340 /F 380 and fura-2 concentrations were considered to be clamped at extracellular [Ca 2+ ] and included in the analysis. Resting [Ca 2+ ] ranged from 40-150 nM in healthy Merkel cells, consistent with resting [Ca 2+ ] in sensory cells [26,27].

Pharmacology
Unless noted, reagents were purchased from Sigma. To assess pharmacological sensitivity of hypotonic-evoked signals, Merkel cells were bathed in modified Ringer's solution containing either 10 mM Ruthenium Red or 50 mM amiloride. At higher concentrations, the fluorescence of amiloride obscured fura-2 fluorescence signals. L-, P/Q-and N-type VACCs were blocked by a mixture of 10 mM nimodipine (Tocris) and 10 mM v-conotoxin MVII-C (Tocris) [8]. Internal Ca 2+ stores were depleted by application of 1 mM thapsigargin (Tocris) followed by repeated high-K + pulses to activate store release.

Volume imaging and analysis
Merkel cells were cultured for two days in an eight-well coverglass chamber (Lab-Tek). Cells were imaged in modified Ringer's solution described above containing 0.1-mm fluorescent microspheres (TetraSpeck, Invitrogen). Microspheres were allowed to settle onto the surfaces of Merkel cells and coverslips for 20-30 min. Merkel cells were imaged with an LSM 5-LIVE imaging system with a Plan-Apocromat 636, 1.4 NA oilimmersion objective lens (Zeiss). Microsphere fluorescence was excited at 532 nm, and epifluorescence emission passed through a 535-nm dichroic beam splitter and a 550-nm long-pass filter. Stacks of confocal sections were imaged once every 7 s. Microsphere locations were determined with the ''spot'' identifying utility in Imaris 5.0.1 software (Bitplane AG). Volume calculations and graphs were generated in custom programs written for MATLAB (Mathworks).

Confocal imaging of Merkel-cell morphology
Cells were imaged in a modified Ringer's solution described above containing 1 mM BODIPY FL C5-ceramide (Molecular Probes) and 0.02% pluronic F-127 (Molecular Probes). Fluorescent sphingolipids were allowed to diffuse into cell membranes for 5 min and then were imaged with the system described for volume imaging.

RT-PCR
GFP + Merkel cells were purified from P3-P6 mice using FACS with strict gating conditions to achieve $95% purity. Cell from two to six mice were used for each sort. Total RNA from 2-10610 4 GFP + cells was isolated using commercially available reagents (Qiagen RNeasy kit) and DNAse treated according to manufacturer's instructions to remove contaminating genomic DNA. First-strand cDNA was synthesized using oligo(dT) [12][13][14][15][16][17][18] primers at 42uC for 2 h with SuperScriptIII (Invitrogen). PCR products were amplified with touchdown PCR using a PTC-200 Peltier thermal cycler (MJ Research); cDNA from ,1000 cells was used for each reaction. To evaluate reproducibility, each primer pair was tested on two to four independent biological samples, that is, cDNA produced from cells purified in separate experiments. We considered amplicons robust if they were present in at least half of biological samples tested. In all experiments, control PCRs lacking cDNA template were performed to confirm the absence of contamination, and primer performance was verified with positive control cDNA template from brain, skin, or a mixture of liver, heart, spleen and kidney tissue. With the exception of PKD-REJ, a gene predicted to have no introns (Entrez Gene accession number NM_011105), primer pairs were designed to span introns to ensure that amplicons were not derived from genomic DNA.

Hypotonic stimuli evoke cytoplasmic Ca 2+ signals in Merkel cells
In cells lacking cell walls, hypotonic extracellular solutions cause water flux across the cell membrane to induce cell swelling. If Merkel cells express ion channels activated by membrane tension, we reasoned that cell swelling might increase membrane tension to open such channels. The resulting membrane depolarization could then activate VACCs to allow Ca 2+ influx. To determine if Merkel cells respond to changes in osmolality, we monitored intracellular Ca 2+ with the ratiometric, fluorescent indicator fura-2. In epidermal-cell suspensions, Merkel cells represent <0.2% of dissociated cells. Using FACS we enriched GFP + Merkel cells to approximately 85% percent: the remaining 15% consisted predominately of GFP-negative keratinocytes. Cells were subjected to Ringer's solutions of varying osmolality. Most Merkel cells showed an increase in free [Ca 2+ ] in response to 20% hypotonic stimuli (6563% cells, N = 19 experiments, 10-33 cells/experiment, Fig. 1A To characterize the dose-response of hypotonic-triggered Ca 2+ increases in Merkel cells, we challenged Merkel cells with solutions of progressively decreasing osmolality. Ringer's solutions of decreasing osmolality induced larger peak Ca 2+ transients in individual Merkel cells (Fig. 1E). Merkel cells with the largest Ca 2+ transients at 10% hypotonic Ringer's solution also had the largest Ca 2+ increase at 20 or 30% hypotonic Ringer's solution. In addition, solutions of progressively lower osmotic strength elicited responses in a greater proportion of Merkel cells than mildly hypotonic solutions (261 mmol?kg 21 recruited 29% of cells, 232 mmol?kg 21 recruited 52% of cells, 203 mmol?kg 21 recruited 71% of cells, Fig. 1F). Our data indicate that Merkel cells respond to relatively mild 10% changes in osmolality, yet, even 30% hypotonic solutions do not appear to saturate Merkel cells' responses.
Hypotonic solutions induce cell swelling that activates stretchsensitive channels in bacteria [28], so we asked whether similar hypotonic-induced swelling occurred in Merkel cells. To ascertain if hypotonic solutions altered Merkel-cell volume, we monitored cell shape in three dimensions with fluorescent microspheres attached to the plasmalemma of Merkel cells while perfusing cells with a 20% hypotonic bath solution. Microspheres settled onto Merkel cells and the surrounding coverslip within 30 min of bath application and remained tightly coupled during solution changes ( Fig. 2A, movie S1). Microspheres were imaged with high-speed confocal microscopy, and their positions were used to model the location of Merkel-cell surfaces in relation to the coverslip. By integrating the volume between reconstructed cell surfaces and the coverslip, we estimated that Merkel cells' volume in isotonic Ringer's solution was 334639 mm 3 (mean6SD, N = 8). To determine if Merkel cells swelled in response to a hypotonic stimulus, we imaged Merkel cells in time series while perfusing with 20% hypotonic Ringer's solution (Fig. 2B). In this condition, Merkel-cell volume was significantly higher (358643 mm 3 , mean6SD, N = 8, p,0.001, paired Student's t test), which represents an average volume increase of 7.362.9%. Merkel cells began swelling within 7 s of the onset of hypotonic perfusion, which was the temporal resolution of the three-dimensional imaging. Merkel cells remained enlarged during 120 s of hypotonic stimulation and relaxed to their original volume after perfusion of isotonic Ringer's solution. One Merkel cell subjected to prolonged hypotonic stimulation showed volume decreases in ,300 s.

Hypotonic-induced Ca 2+ transients are concentrated in processes
Mechanosensitive ion channels are often located within specialized cellular processes that are thought to leverage forces to the channels. Hair cells have mechanoelectrical transduction channels near the tips of modified microvilli called stereocilia [29], and kidney cells detect fluid flow with mechanosensitive channels located in cilia [30]. Similarly, Merkel cells in vivo have actin-filled processes that penetrate overlying keratinocytes [5,31]. To determine if Merkel cells extend cytoplasmic processes in vitro, we stained Merkel cells with fluorescent sphingolipids to visualize membrane morphology (Fig. 2C). We found that most cultured Merkel cells have processes arranged in a branch-like pattern, with numerous fine protrusions (1-8 mm in length) at the terminals of large processes (2-15 mm in length). Although the larger cytoplasmic processes were visible in fura-2 fluorescence images, fine protrusions were not (Figs. 1, 3).
During hypotonic stimuli, elevated [Ca 2+ ] was evident in Merkel-cell cytoplasmic processes and around nuclei (Fig. 3A, B, movie S2). In cytoplasmic processes, [Ca 2+ ] increased first in regions adjacent to the plasmalemma and then in regions located deeper within the cytoplasm (Fig. 3C, D) Fig. 4A, E), which recovered upon reintroduction of external Ca 2+ . These data indicate that extracellular Ca 2+ influx is required for hypotonic-triggered Ca 2+ transients.
To ascertain whether VACCs also contribute to hypotonicinduced increases in free [Ca 2+ ], we blocked these channels with a cocktail containing 10 mM conotoxin MVIIC to block N-and P/ Q-type Ca 2+ channels and 10 mM nimodipine to block L-type Ca 2+ channels [8]. The efficacy of this cocktail was tested by depolarizing Merkel cells with high-K + Ringer's solution (Fig. 4B). The blocking cocktail inhibited peak high-K + -induced Ca 2+ transients by 9661% (mean6SEM, N = 4 experiments). By contrast, the VACC-inhibitor cocktail only curtailed peak hypotonic-induced Ca 2+ transients by 51613% (mean6SEM, N = 4 experiments). Thus, in the presence of VGCC blockers, 60% of Merkel cells had larger hypotonic-induced transients than high-K + induced transients (Fig. 4B). This partial inhibition of hypotonic-evoked Ca 2+ signals by VACC blockers indicates that VACCs contribute to, but are not the sole source of, hypotonicinduced Ca 2+ influx across the plasma membrane.
To determine whether internal Ca 2+ stores add to the hypotonic-induced Ca 2+ increases, we designed an experimental protocol to empty internal Ca 2+ stores (Fig. 4C, D). We blocked Ca 2+ reuptake into internal stores with 1 mM thapsigargin, which inhibits the sarco/endoplasmic reticulum Ca 2+ -ATPase. As expected, Merkel cells displayed an increase in cytoplasmic Ca 2+ upon thapsigargin treatment (data not shown and Piskorowski et al., submitted). To ensure that Ca 2+ stores were depleted after thapsigargin treatment, we depolarized cells with high-K + Ringer's solution to activate Ca 2+ -induced Ca 2+ -release. Store depletion was verified at the end of experiments by chelating extracellular Ca 2+ with EGTA and permeabilizing cell membranes with the Ca 2+ ionophore ionomycin. In the absence of thapsigargin,  ionomycin elicited cytoplasmic Ca 2+ transients (Fig. 4D), demonstrating that Merkel cells have detectable Ca 2+ stores. By contrast, ionomycin-induced transients were reduced by 9261% in cell pretreated with thapsigargin (Fig. 4C, mean6SEM, N = 31 cells). Pre-treatment with thapsigargin also reduced Ca 2+ transients evoked by hypotonic stimuli by 8063% (mean6SEM, N = 4 experiments).
Together, these data demonstrate that the amplitudes of hypotonic-induced transients in the presence of thapsigargin and VACC antagonists, though reduced, were significantly larger than those in the presence of 10 mM EGTA (p,0.05, paired Student's t test, N = 3-4 experiments, Fig. 4E). These results indicate that internal Ca 2+ stores augment Ca 2+ influx induced by hypotonic solutions.

Pharmacology of hypotonic-evoked Ca 2+ signals
Hypotonic induced extracellular Ca 2+ influx implies the presence of a swelling-activated ion channel in the plasma membrane but does not speak to its identity. The most obvious candidate is TRPV4 (GenBank accession number NM_022017), which is expressed in Merkel cells [32] and is activated by hypotonic solutions when expressed in human embryonic kidney (HEK) cells [33,34]. To determine if TRPV4 is required for hypotonic activation, we analyzed hypotonic-induced Ca 2+ transients in Merkel cells from TRPV4-deficient mice (Fig. 5). The magnitude and time course of Ca 2+ transients in Merkel cells from TRPV4 -/-mice were indistinguishable from those of heterozygous littermate controls (Fig. 5) and from wild-type responses. Thus, TRPV4 is unlikely to mediate hypotonic induced Ca 2+ influx in Merkel cells.
We next broadened our search for the identity of swellingactivated channels in Merkel cells. Two ion-channel families have been implicated in mechanotransduction in mammals: transient receptor potential (TRP) channels and Degenerin/Epithelial Na + channels (DEG/ENaC), [35]. Ruthenium red is a broad-spectrum blocker of many channels, including TRPV channels. We found . Hypotonic responses of Merkel cells exposed to these compounds were normalized to control hypotonic responses. Error bars indicate SEM. Asterisks denote statistically significant differences between EGTA and VACC or thapsigargin treated responses (p#0.04, paired Student's t test). The un-normalized EGTA, VACC and thapsigargin treated responses were significantly different from their respective controls (p#0.05, Student's t test). doi:10.1371/journal.pone.0001750.g004 that 10 mM ruthenium red inhibited peak hypotonic-induced Ca 2+ transients by 7168% (mean6SEM, N = 2 experiments, N = 9-10 cells/experiment); however, ruthenium red inhibited high-K +induced Ca 2+ transients to the same extent (68612%). Because ruthenium red did not fully block hypotonic-induced Ca 2+ transients, we sought to determine if hypotonic-evoked Ca 2+ signals were sensitive to the DEG/ENaC antagonist, amiloride. Amiloride (50 mM) did not inhibit the hypotonic-induced Ca 2+ transient (28613%, N = 2 experiments, N = 20-27 cells/experiment).

Merkel cells express TRP channels
Since amiloride did not inhibit the hypotonic-evoked Ca 2+ increases, we used RT-PCR to screen for amiloride-insensitive non-selective cation channels expressed in Merkel cells. We focused on channels from the TRP family because many of these channels fit this profile and because they function in diverse modes of sensory transduction. Primer efficacy was tested against cDNA derived from brain, skin, or a mixture of liver, skin, heart, spleen and kidney (Fig. 6). Merkel-cell cDNA yielded robust amplicons for six TRP channels and occasional amplicons for an additional five channels (Table 1). Notably, we detected amplicons for TRPC1 (GenBank accession number NM_011643), PKD1 (NM_013630) and PKD2 (NM_008861), channels previously implicated in mechanotransduction in other cell types [36,37].

Discussion
This study demonstrates that dissociated Merkel cells are directly activated by hypotonic-evoked cell swelling, and introduces a robust in vitro assay for analyzing the molecular mechanisms that underlie swelling-evoked signals. Based on our findings, we propose that swelling triggers Ca 2+ entry through an as yet unknown cation channel; the resultant depolarization activates VACCs and together these two sources of Ca 2+ influx activate Ca 2+ release from internal stores (Fig. 7). We identified 11 TRP channels expressed in Merkel-cells, seven of which have pharmacological profiles matching the hypotonic-evoked Ca 2+ responses we observed. Of these, PKD1 and PKD2 are promising candidates because they have been previously implicated in mechanotransduction in cilia of kidney cells [36]. Our data support a model in which skin indentation applies force to Merkel cells, whose mechanosensitive channels allow Ca 2+ influx. This Ca 2+ influx is augmented by VACCs and Ca 2+ release from internal stores. These Ca 2+ increases could trigger synaptic signaling to the underlying sensory afferent.

Merkel cells express hypotonic-activated ion channels
Several lines of evidence indicate that Ca 2+ -permeable ion channels generate hypotonic-induced Ca 2+ transients in Merkel cells. First, the requirement for extracellular Ca 2+ suggests Ca 2+ ions enter the cell across the plasma membrane. Second, blocking VACCs or emptying intracellular stores curtails, but does not eliminate, hypotonic induced Ca 2+ increases, suggesting other sources of Ca 2+ entry (Fig. 7). Although it is formally possible that a hypotonic-activated G-protein coupled receptor could induce membrane depolarization and Ca 2+ release from stores, such a hypotonic-activated receptor would have to be inactivated by extracellular EGTA to explain our findings. We know of no receptors that match these requirements. Thus, the most parsimonious interpretation of our data is that hypotonic solutions induce Merkel cells to swell, causing increased membrane tension that activates Ca 2+ -permeable ion channels (Fig. 7).
Does osmosensitivity imply that Merkel cells are mechanosensory cells? Most cells have volume-regulating mechanisms that permit adaptation to osmotic changes in their environment. In addition, a variety of excitable and non-excitable cells have detectable Ca 2+ increases in anisotonic extracellular solutions [38][39][40]. Two lines of evidence suggest hypotonic solutions might activate mechanotransduction machinery rather than ubiquitous volume regulating pathways. First, in some non-excitable cells, regulatory volume decrease is preceded by hypotonic-evoked Ca 2+ transients with 1-5 min latencies followed by 2-10 min time to peak. By contrast, a subset of sensory cells isolated from the trigeminal nucleus has robust hypotonoic-evoked responses that develop within seconds [24]. These fast responding neurons have been proposed to be a mechanosensitive subset of trigeminal neurons. We found that hypotonic solutions induced a Ca 2+ signal in Merkel cells similar to this rapidly activating neuronal population. Furthermore, these Ca 2+ signals were not temporally correlated with decreases in cell volume, as would be expected for regulatory volume decreases.
The latency of hypotonic-induced Ca 2+ influx in Merkel cells in vitro is <11 s, which is much longer than the 200 ms latency of the SAI response [41]. One explanation for this difference is that distinct molecular mechanisms may transduce hypotonic stimuli in vitro and touch in vivo. Alternatively, it is possible that the same mechanotransduction molecules are differentially activated in these two contexts. For example, Merkel cells may respond rapidly to touch-evoked pressure, whereas osmotic stimuli might take longer to develop sufficient membrane distortion to activate channels. Although our volumetric data indicate that Merkel cells begin to swell within 7 s of hypotonic-solution perfusion, it may   Table 1. Bars at base of figure mark lanes with products amplified from GFP + Merkel cells or control cDNA, respectively. Control cDNA was derived from brain or skin. doi:10.1371/journal.pone.0001750.g006 take the observed 11 s to generate sufficient membrane tension to activate mechanotransduction channels. Alternatively, optimal gating of the Merkel cells' transduction channels in vivo may require extracellular linkages that are not present in culture. Extracellular links are required to transmit force to mechanotransduction channels in hair cells [35]. Moreover, mutations in specific extracellular molecules disrupt touch responses in Caenorhabditis elegans [42]. We postulated that hypotonic-induced membrane tension constitutes a global mechanical stimulus that is sufficient to activate force-transducing machinery in Merkel cells, but is less efficient than touch-evoked channel gating in vivo. The lack of extracellular linkages might also explain why direct touch failed to generate membrane currents in isolated Merkel cells in a previous report [12]. Finally, the latency of mechanotransduction is unknown in Merkel cells in vivo. Although the SAI response has a latency of 200 ms, it is possible that the initial phase of the SAI response is generated by the afferent and the static phase is transduced by Merkel cells [43].
In vivo, Merkel cells' superficial surfaces are studded with dozens of actin-rich microvilli that are proposed to be sites of mechanotransduction [9,31]. In culture, we observed Merkel cells with two basic morphologies: cells with large processes 2-15 mm in length that branch into smaller protrusions and spherical cells with only the smaller protrusions. Both cell shapes are observed in vivo, where they are termed dendritic and non-dendritic Merkel cells [44][45][46]. In dendritic Merkel cells in vitro, we often observed hypotonic-triggered Ca 2+ transients in processes before global cytoplasmic Ca 2+ levels increased. This might reflect Ca 2+ influx into the confined cytoplasmic volume of the process or local differences in Ca 2+ regulation. Alternatively, these data could indicate a higher density of osmotically activated channels in these processes.

Candidate transduction channels in Merkel cells
Several gene families have been implicated in vertebrate mechanotransduction, including members of the TRP family and the DEG/ENaC family. In particular, TRPV4 responds to hypotonic solutions in HEK cells [33,34] and is expressed in Merkel cells [32]. Somewhat surprisingly, our analysis TRPV4deficient mice indicates that this channel is not required for swelling-activated Ca 2+ signals in Merkel cells. Thus, we used pharmacology to elucidate the molecular nature of Merkel cells' hypotonic-activated channel. Amiloride (50 mM), which inhibits most DEG/ENaC channels with an IC 50 of 0.1-20 mM [47], does not block hypotonic induced Ca 2+ signals, making these channels unlikely candidates for Merkel-cell transduction channels. Ruthenium red is a broad-spectrum inhibitor that blocks some TRP channels, including TRPV channels. We found that 10 mM ruthenium red inhibits both hypotonic and high-K + induced Ca 2+ transients in Merkel cells. Since ruthenium red has also been shown to inhibit both VACCs and Ca 2+ -store release in some cell types [48,49], it is unclear to what extent ruthenium red directly inhibits swelling-activated channels or downstream Ca 2+ signaling. At any rate, because ruthenium red does not completely block hypotonic-induced Ca 2+ transients, they cannot be solely mediated by ruthenium-red-sensitive channels, including the TRPV subfamily.
Because other TRP channels play prominent roles in sensory transduction, including mechanotransduction in zebrafish and invertebrates, they remain attractive candidates for investigation. Moreover, many of these channels are not blocked by ruthenium red. We demonstrated with RT-PCR that Merkel cells express robust transcripts encoding six TRP channels, and transcripts were occasionally amplified for an additional five channels. The sporadic amplification of these latter transcripts may indicate that they are present in very low copy number [50].
Among the TRP channels expressed in Merkel cells, TRPC1, PKD1 and PKD2 have been previously implicated in mechanotransduction. Both are Ca 2+ permeable, and neither is known to be blocked by ruthenium red [36,51,52]. Although PKD2 is blocked by amiloride, its IC 50 is 80 mM, larger than the concentrations than we tested here (see Methods and [53]). Interestingly, PKD1 and PKD2 form Ca 2+ permeable complexes that transduce fluid flow in the kidney [30]. In human polycystic kidney disease, patients are heterozygous for either PKD1 or PKD2 mutations and suffer from cyst formation and eventual kidney failure [54]. As homozygous deletion in PKD1 or PKD2 in mice causes embryonic death, examination of touch sensitivity in PKD-deficient mice will require a conditional knockout [55]. TRPC1 is activated by membrane tension in CHO-cells and Xenopus laevis oocytes [37]; however, the role of TRPC1 in vivo is still undetermined [56]. Many of the TRP channels we found expressed in Merkel cells are orphans: little is known of their biophysical properties, let alone their physiological roles in vivo. Consequently, they represent candidate transduction channels. Thus, our findings set the stage for future gene disruption experiments to identify which of these channels mediate swelling-activated signals in Merkel cells in vitro and to determine whether the same channels transduce touch stimuli in vivo.

Supporting Information
Movie S1 Time series of confocal z-series projections of a Merkel cell during a 20% hypotonic perfusion. The cell surface was visualized with fluorescent microspheres (white spots). Microspheres on the coverslip surface have been removed for clarity. Z-series were collected every 7 s.