Merkel Cells as Putative Regulatory Cells in Skin Disorders: An In Vitro Study

Merkel cells (MCs) are involved in mechanoreception, but several lines of evidence suggest that they may also participate in skin disorders through the release of neuropeptides and hormones. In addition, MC hyperplasias have been reported in inflammatory skin diseases. However, neither proliferation nor reactions to the epidermal environment have been demonstrated. We established a culture model enriched in swine MCs to analyze their proliferative capability and to discover MC survival factors and modulators of MC neuroendocrine properties. In culture, MCs reacted to bFGF by extending outgrowths. Conversely, neurotrophins failed to induce cell spreading, suggesting that they do not act as a growth factor for MCs. For the first time, we provide evidence of proliferation in culture through Ki-67 immunoreactivity. We also found that MCs reacted to histamine or activation of the proton gated/osmoreceptor TRPV4 by releasing vasoactive intestinal peptide (VIP). Since VIP is involved in many pathophysiological processes, its release suggests a putative regulatory role for MCs in skin disorders. Moreover, in contrast to mechanotransduction, neuropeptide exocytosis was Ca2+-independent, as inhibition of Ca2+ channels or culture in the absence of Ca2+ failed to decrease the amount of VIP released. We conclude that neuropeptide release and neurotransmitter exocytosis may be two distinct pathways that are differentially regulated.


Introduction
Merkel cells (MCs) are cutaneous neuroendocrine cells that are mainly found in touch-sensitive areas of the glabrous epidermis and outer root sheaths of hair follicles [1,2,3]. MCs are discerned from keratinocytes by their close connections to sensory neurons, the presence of dense-core neurosecretory granules and expression of cytokeratin (CK) 20. Their relationship with nerve terminals and their particular location have led to the hypothesis that MCs function in mechanoreception [4]. Furthermore, convincing evidences have demonstrated their excitability [5], expression of fundamental proteins of the synaptic machinery [6], synaptic-like contact with sensory neurons [7,8] and involvement of the Ca 2+induced Ca 2+ release (CICR) pathway in the production of Ca 2+ transients [9,10,11].
In addition, MCs produce bioactive amines and peptides such as serotonin, met-enkephalin, chromogranin A, calcitonin generelated peptide (CGRP) and vasoactive intestinal peptide (VIP), which are held in dense-core granules [12,13]. Thus, they belong to the neuroendocrine cell family. It is widely accepted that neuroendocrine cells participate in tissue homeostasis, cell growth and regeneration. The presence of non-innervated MCs in hair follicles [14] and in oral mucosa [15] fosters the idea that MCs assume such functions. Finally, MCs are thought to stimulate keratinocyte proliferation, to regulate skin homeostasis and the hair cycle and to play a trophic role toward sensory neurons [16,17]. Unfortunately, the role of MCs in the cutaneous environment remains largely unexplored.
MCs appear to be quite scarce in the epidermis, which has hampered their study in vitro. Analyses performed in total epidermal cell cultures revealed that they are difficult to maintain for more than two weeks [18,19]. MCs were not labeled by BrdU [20], they did not incorporate 3 H-thymidine [21] and they were not immunoreactive to Ki-67 [22,23], an antigen that is associated with proliferating cells. Thus, in contrast to neighboring keratinocytes, they apparently do not renew. However, MC hyperplasias have been reported in some inflammatory skin diseases [24], such as actinic keratosis [25], neurodermatitis [26] and fibrous papules [27], which suggests that they proliferate under certain conditions and that they have a role in skin disorders.
In the skin, VIP is produced by MCs and some peptidergic nerve terminals [28,29]. VIP was shown to enhance the proliferation of keratinocytes [30], to induce mast cell degranulation [31] and to enhance IL-1a, IL-6, IL-8, RANTES and TNFa expression by keratinocytes [32] [33]. Conversely, VIP exhibits immuno-modulatory activity by decreasing the production of proinflammatory cytokines [34,35]. Since VIP expression is upregulated during psoriasis [36], atopic dermatitis [37] (with conflicting results) and skin disorders like aquadynia [38], better knowledge of the mediators leading to VIP release by MCs is of great interest for the understanding of these diseases.
To gain further insights into the neuroendocrine functions of MCs, we developed cultures enriched in MCs from swine snouts. This cell type is enriched in the swine snout. Different growth factors and molecules were compared. We show here that, in the presence of serum and a low proportion of keratinocytes, MCs were able to proliferate. In culture, MCs reacted to histamine and were able to sense tissue acidification and cell swelling through the transient receptor potential vanilloid (TRPV) type 4 receptor. These conditions induced the release of VIP. Conversely, acetylcholine (Ach) inhibited neuropeptide exocytosis. Moreover, in contrast to the dependence of their mechanotransduction properties on Ca 2+ signaling, neuropeptide release involved a (Ca 2+ )-independent secretory pathway that was inhibited by Ca 2+ .

Enrichment for MCs
MCs are rare neuroendocrine cells of the epidermis. Due to their rarity, a great challenge to their study is the harvest of a sufficient number of cells to perform experiments. A significant advance was achieved by the purification of GFP-positive MCs from transgenic mice [6,39], but this method remains restrictive for most laboratories. Here, we used a positive magnetic cell sorting strategy to enrich for MCs. The swine snout was preferred to human biopsies for its higher proportion of MCs as demonstrated by CK20 immunostaining (Figure 1).
An average of 45.7 (613)610 6 dissociated epidermal cells was obtained per swine snout. We targeted CD56 antigen for the purpose of cell sorting. It is one of the only extracellular markers described for human and swine MCs [40,41], although all MCs do not express it [42]. We routinely obtained ,5.9610 5 cells (1.5460.15% of the total dissociated cells) in the CD56-positive fraction, of which MCs accounted for an average of 62% (63.5) as shown in over 10 experiments employing CK20 immunofluorescence and DAPI counterstaining (Figure 2a). In contrast, CK20positive cells were rarely observed in the unlabeled fraction. To confirm this result, collected cells were post-fixed in glutaraldehyde buffer and observed by electron microscopy. The ultrastructural analysis confirmed the predominance of small cells with large, polylobulated nuclei and clear cytoplasm containing dense-core granules of an average diameter of 100 nm ( Figure 3). This ultrastructure is characteristic of MCs [1,43], which demonstrates the ability of this method to enrich for MCs.

Merkel cells in culture
Culture of MCs has been reported to be very difficult. Under previously reported culture conditions, cells underwent apoptosis within two weeks and no proliferation was observed [7,20]. However, MC hyperplasia has been reported; therefore, MCs can proliferate in response to specific factors. We exposed cultured MCs to different media, growth factors and mediators. The medium was changed every two days. MCs were analyzed after three days by immunofluorescence. The keratinocyte growth medium (K-SFM) did not allow MCs to develop cytoplasmic extensions. Furthermore, this culture method led to a rapid decrease in the proportion of MCs due to keratinocyte proliferation. Therefore, DMEM/F12 was used as a basal medium. The addition of EGF (20 ng/mL) also failed to affect MCs, as they remained round in shape. This result was consistent with their neural crest origin [44]. However, none of NGF (100 ng/mL), NT-3 (25 ng/mL), BDNF (25 ng/mL) or all three together modified MC behavior (Figure 2b). Similarly, MCs were not affected by RA (0.5 mM).
MC hyperplasia occurs in some inflammatory diseases, but culture with IL-6 or TNF-a did not lead to MC proliferation [7]. Hence, we cultured MCs with glucocorticoid (Dexa, 1 mM) to evoke growth arrest and differentiation with the aim of observing cell spreading. However, Dexa did not induce extension of cytoplasmic outgrowths. By contrast, once stimulated by bFGF (20 ng/mL), MCs clearly developed cytoplasmic processes and became easily distinguishable from keratinocytes ( Figure 2c). These morphological changes reflected the viability of the cells and indicated that bFGF is a likely survival factor for MCs.

Merkel cells can proliferate in vitro
Finally, we cultured MCs in DMEM/F12 serum-supplemented medium (7% FCS and 7% HS). After three days, most of the cells assumed a dendritic shape more readily than with bFGF. In this condition, MCs established numerous connections between one another, and several stacks of dendritic MCs were observed (Figure 4a, b). This particular configuration was confirmed by immunofluorescence against CK20 ( Figure 4c). Strikingly, this particular arrangement evoked the configuration that MCs physiologically take in areas sensitive to touch, like in the footpads of rodents, as we found by immunohistochemistry of dissociated epidermal layers ( Figure 4d). Therefore, we addressed the question of how such arrangements develop. Specific cell migrations were contested and MCs are thought to be terminally differentiated so they may not proliferate [7,22,23]. In the present study, we tested the immunoreactivity of cultured MCs to the proliferation marker Ki-67. Unexpectedly, many cells were positive for both CK20 and Ki-67.All phases of mitosis were visible: prophase (Figure 5a (Figure 5d, e). MCs were followed for over four weeks in culture, but proliferating CK20positive cells were detected only during the first two weeks after enrichment. These results demonstrate for the first time that MCs can undergo mitosis in vitro. We supposed that keratinocyte factors disable this capability, which would explain why previous reports failed to amplify MCs.

MCs produce VIP in culture
MCs of most mammalian species produce VIP [45]. Conversely, keratinocytes do not express it under physiological conditions [38]. In our culture conditions, MCs were immunoreactive to VIP as revealed by double immunofluorescence ( Figure 6). The merge picture demonstrated that nearly all CK20-positive MCs also produced VIP, and that cells other than MCs were not found to express VIP. In addition, the synthesis of VIP was detected by western blot (WB) analysis of whole cell homogenates produced from cultures enriched in MCs ( Figure 6, WB). In contrast, calcitonin gene-related peptide (CGRP) and substance P were produced by MCs and keratinocytes. Neuropeptide Y (NPY) was hardly detectable in our cultures (data not shown).

Ca 2+ signaling in neuropeptide release from MCs
MCs were shown to express the histamine H3 receptor [46] and TRPV4 [47], and Ca 2+ signaling was demonstrated to be an essential component in mechanoreception [9]. We analyzed how these factors modulate neuropeptide exocytosis from MCs by monitoring the amount of VIP released into the culture medium by ELISA. Cells were stimulated in DMEM/F12, which contains 1 mM Ca 2+ , as a control condition. Supernatant was harvested and analyzed after 5 min. The basal level was defined as 100%.
To study the involvement of Ca 2+ signaling in neuropeptide release, a 5-min pre-incubation was performed with the Ca 2+ channel inhibitor ruthenium red (RR) (1 mM). Unexpectedly, this treatment increased the VIP concentration in the supernatant following histamine (+72%) as well as to 4aPDD (+70%) stimulation. To confirm this result, we stimulated MCs by treatment with 4aPDD in Ca 2+ -free PBS. In this case, a dramatic enhancement of VIP release was measured (+131%). These results strongly suggest that dense-core granule exocytosis occurs via Ca 2+ -independent signaling. Moreover, activation of Ca 2+ signaling seems to interfere with VIP release.
In some neurons and neuroendocrine cells in which both densecore granules and synaptic vesicles co-exist, an intracellular pathway other than Ca 2+ signaling can modulate exocytosis [48]. This pathway exists in neurons that co-produce Ach as a neurotransmitter and neuropeptides such as VIP. In such a configuration, Ach has been reported to inhibit VIP release through the muscarinic Ach receptor [49,50]. To find out whether MCs respond like these cells, we stimulated them with Ach. Addition of Ach (10 mM) to the culture medium reduced the release of VIP by 23%; however, pre-inhibition of Ca 2+ channels with RR (1 mM) before Ach stimulation enhanced VIP release by 107% (Figure 7, n = 4). These results are consistent with signaling via the muscarinic Ach receptor.
To explore this phenomenon further, we stimulated MCs in Ca 2+ -free medium (MEME). In this condition, Ach failed to inhibit VIP release significantly (Figure 8, P.0.1, representative results from two experiments). The addition of Ca 2+ (1 mM) to the culture medium was sufficient to reduce this amount to a lower level (216%), which corroborates the previous results (Figure 7 and 8). Stimulation by Ach in the presence of Ca 2+ markedly inhibited VIP exocytosis (231%), thereby confirming the involvement of Ca 2+ signaling in the inhibition of neuropeptide release.

Discussion
Consistent with their neural crest origin [44], MCs were not found to renew like keratinocytes [20,21,22]. Thus, they are believed to have a long life span, similar to neurons. MCs act in mechanoreception, but their exact role remains to be identified. They are thought to participate in skin homeostasis through the release of bioactive molecules, but few reports have focused on the neuroendocrine properties of MCs. This lack of knowledge is mainly due to the lack of an experimental model. We developed a method for generating cultures enriched in porcine MCs. In the process of this work, we identified in vitro survival factors and demonstrated that MCs are able to proliferate in culture. Subsequently, we found that the MCs reacted to histamine and activation of TRPV4. These stimulations led to VIP release in a Ca 2+ -independent manner. Conversely, Ach inhibited VIP release, probably through the activation of Ca 2+ channels.
Previous attempts to culture MCs were performed in total epidermal cell cultures in which keratinocytes were predominant. In these cultures, MCs were difficult to maintain for more than two weeks, they did not undergo mitosis and the low number of MCs restricted the opportunities for analysis [7,18,51]. To overcome these obstacles, E. A. Lumpkin and her team used transgenic mice from which GFP-positive MCs were efficiently selected by fluorescence-activated cell sorting (FACS) [6]. Here, we used a magnetic cell sorting system to select MCs from swine snout, as this tissue has one of the highest densities of MCs among mammalian tissues (Figure 1). Compared to FACS selection, we achieved a lower ratio of MCs in our cultures (60-65% versus 85-95%), but a higher number of MCs per animal. In addition, this technique has been successfully used to enrich human MCs, but with a yield proportional to the lower initial MC density in this tissue.
Through morphological analysis by fluorescence microscopy, we observed MC spreading in the presence of bFGF. The extension of outgrowths reflected the activity of a growth factor. This finding suggests involvement of the bFGF pathway in MC hyperplasia and possibly in Merkel cell carcinoma (MCC). By contrast, keratinocyte growth factors did not stimulate the development of MC extensions. This finding could be consistent with their neural crest origin; however, all of the tested neurotrophins failed to support the development of cytoplasmic extensions (Figure 2b). This result is surprising given that the nerve dependency of MCs in vivo has been reported several times. First, mice lacking receptors for NGF, BDNF and NT-3 (TrkA, TrkB and TrkC respectively) have a lower number of MCs than control mice [52,53]. Also, NT-3 appears not to be essential for the development of MCs, but its absence leads to perinatal apoptosis of MCs [54]. Finally, overexpression of NT-3 and BDNF increased the number of MCs [55,56]. In fact, all these mutations also affect the cutaneous nerve density, which correlates with MC number. Our in vitro results seem to demonstrate that neurotrophins alone are not sufficient to maintain MCs in culture. Hence, other paracrine factors from sensory nerve endings probably assume this function.
When cultured in serum-supplemented medium, alignment of MCs was observed (Figure 4a, b). To the best of our knowledge, this configuration has not been reported before. Interestingly, in vitro staking of MCs can be compared to the in vivo staking of MCs detected in rodent touch pads (Figure 4d), and perhaps they imply the same events. The proliferative ability of MCs has been a matter of debate for decades. Based on Ki-67 immunoreactivity, we provide direct evidence that MCs can undergo mitosis in vitro ( Figure 5), although most studies have concluded that they are terminally differentiated. We hypothesize that the predominance of epidermal factors regulates this capability, as suggested by our results (see above). Thus, an environment rich in MC growth factors, as occurs in MC hyperplasia, could support the proliferation of MCs. In our model, MCs underwent mitosis during the first two weeks, before overgrowth by keratinocytes. Further analysis of the signaling pathway that initiates primary MC mitosis would be of interest in the study of MCC. The recently identified MC polyomavirus [57] integrated in most MCCs probably interferes with this pathway.
We demonstrated here that MCs react to histamine and to TRPV4 stimulation by releasing VIP (Figure 7). Thus, in the skin, MCs would be able to respond to mast cell degranulation, cell swelling and tissue acidification. Since VIP released in the epidermal environment is known to have immuno-modulatory   functions and effects on keratinocytes, our findings support a regulatory role of MCs in skin diseases. VIP differentially regulates immune functions according to the tissue and the disease, and its role in skin disorders has yet to be detailed. To our knowledge, these data are the first to support a participatory role of MCs in skin pathophysiology.
The level of released VIP was lower for cells cultured in DMEM:F12 (Ca 2+ , 1 mM) compared to those cultured in PBS without Ca 2+ (Figure 7). In order to avoid interactions with components of the culture medium that are not present in PBS, we compared the amounts of VIP released in MEME (a medium without Ca 2+ ) with or without the addition of Ca 2+ (Figure 8), and we found similar results. These data contrast with the Ca 2+dependence of the synaptic vesicle exocytosis underlying mechanotransduction. Thus, two distinct secretory pathways are indicat-  . Exposure to the TRPV4 agonist 4aPDD (1 mM) or histamine (100 mM) stimulated VIP release from MCs. This release was not inhibited by pre-incubation with the Ca 2+ channel inhibitor RR (1 mM), indicating a Ca 2+ -independent pathway. The marked increase of VIP release obtained following stimulation by 4aPDD in a Ca 2+ -free buffer confirmed these results. In addition, exposure to the neurotransmitter Ach (10 mM) significantly inhibited the amount of VIP released in the presence of Ca 2+ . This inhibition was suppressed by preincubation with RR (1 mM), suggesting the involvement of Ca 2+ channels (n = 4, difference significant at P#0.025.). doi:10.1371/journal.pone.0006528.g007 Figure 8. Ca 2+ and Ach decreased the release of VIP. Effect of Ach and Ca 2+ on VIP release expressed in ng/well. MCs were cultured in Ca 2+ -free MEME and then exposed to Ach (100 mM), Ca 2+ (1 mM), or both. In the absence of Ca 2+ , the addition of Ach failed to significantly decrease VIP exocytosis (P.0.1). A lower amount of VIP release was observed in the presence of Ca 2+ (1 mM). Finally, the inhibitory effect of Ach on VIP release required extracellular Ca 2+ (P#0.025). These results are consistent with inhibition via Ca 2+ channels, probably via muscarinic M2 or M4 Ach receptors. doi:10.1371/journal.pone.0006528.g008 ed. First, a Ca 2+ -dependent signaling pathway activates the CICR pathway, which leads to neurotransmitter release in the mechanotransduction process [9,10,11]. Second, a Ca 2+ -independent pathway is involved in dense-core granule exocytosis. Our data support the hypothesis that activation of Ca 2+ -dependent signaling inhibits the Ca 2+ -independent pathway involved in neuropeptide release, since inhibition of the first one increased the level of activation of the second one.
In neurons and neuroendocrine cells where two secretory pathway co-exist, exocytosis could be regulated or modulated by intracellular signals such as cAMP or diacylglycerol [49]. This pathway remains to be clarified for neuropeptide release from MCs. Furthermore, Consistent with previous studies that have described interference between Ach exocytosis as a neurotransmitter and VIP release [50], we demonstrated in MCs that Ach inhibits VIP release in a Ca 2+ -dependent manner, probably through M2 or M4 muscarinic Ach autoreceptors [49]. Based on these results, we propose Ach as a putative neurotransmitter for sensory nerve endings, since met-enkephalin and glutamate did not satisfy this function [58,59].
In conclusion, this work describes a new way to culture of MCs. In our experimental model, we provide evidence that MCs can undergo mitosis. Through the release of VIP following histamine stimulation and activation of TRPV4, MCs probably act in skin pathophysiology, but their exact role remains to be defined. The proliferation of MCs in particular conditions could amplify these effects. In addition, the mechanotransduction properties of MCs appear to be independent of neuropeptide exocytosis, as the latter does not require Ca 2+ signaling. Finally, we characterized the role of Ach in neurosecretory release.

Immunohistochemistry
Pieces of skin biopsies were embedded in OCT, cryopreserved in isopentane chilled on liquid nitrogen and sections with a thickness of 10 mm were cut. Slides were saturated with 5% NGS in PBS with 0.05% Triton X-100 for 15 minutes and subsequently hybridized with primary antibodies. Slides were rinsed twice and hybridized with appropriate secondary antibodies for two hours at room temperature. The antibodies used were those described above.

Electron microscopy
MC-enriched suspensions were fixed in 2.5% glutaraldehyde for 30 minutes, rinsed in 0.1 M cacodylate buffer, and post-fixed for one hour with 1% OsO 4 . Cells were dehydrated in graded concentrations of alcohol, and finally by propylene oxide. Then, cells were embedded in epoxy resin for one day at 60uC. Finally, ultrathin sections were cut and observed with a Jeol, Jem-1010 Electron Microscope.

Analysis of neuropeptide release
VIP is produced by MCs of most mammalian species [28]. Its release was studied by western blot (WB) and ELISA. MCs were seeded at 5610 4 cells per well in 24-well tissue culture plates for three days in growth medium. Cells were washed twice and then exposed to histamine (100 mM), Ach (10 mM), 4 aPDD (1 mM) or Ca 2+ (1 mM) in DMEM-F12, PBS or MEME with or without Ca 2+ . In some cases, cells were pre-exposed for 5 minutes to the Ca 2+ channel antagonist ruthenium red (1 mM). All chemicals were purchased from Sigma-Aldrich. After a 5-minute incubation, culture supernatants were collected and centrifuged to remove detached cells. For WB analysis, supernatants were diluted in Laemmli buffer (100 mM Tris/HCl, 4% SDS, 15% glycerol, 15% 2-mercaptoethanol, 20 mg/mL bromophenol blue). Proteins were separated by 12% SDS-PAGE and hybridized to nitrocellulose membranes (PALL). The membranes were proved with the indicated antibodies. Proteins were detected by chemiluminescence using the Western Blotting Luminol Reagent (Santa Cruz). For ELISA, 50 mL of supernatant were assessed in triplicate in 96well ELISA microplates (Greiner Bio One). The wells were saturated with 3% BSA in PBS-T over-night in order to allow for complete binding. Hybridizations with primary and secondary antibodies were carried out for two hours in PBS-T 0.3% BSA. The amount of neuropeptide was assessed using the SureBlue Reserve TMB Microwell Peroxidase substrate (KPL). To measure the amount of VIP release, the blocking peptide VIP (Santa Cruz) was used as a standard. A nonparametric Mann-Whitney test was used for statistical analyses.

Author Contributions
Conceived and designed the experiments: NB NR CC LM. Performed the experiments: NB UP NL EG LM. Analyzed the data: NB UP NL EG NR CC LM. Contributed reagents/materials/analysis tools: NB UP NL EG GD LM. Wrote the paper: NB LM.