Cell-cell signaling elicits local Ca2+ transients in melanocyte dendrites and dendritic spine-like structures

Compartmentalized cytoplasmic fluctuations of Ca2+ within dendrites and dendritic spines regulate a variety of neuronal functions. Like some neurons and glia, melanocytes are neural crest derived and possess dendrites (Adameyko et al., 2009; Erickson et al., 1992; Fitzpatrick and Szabo, 1959). Here, we show that primary human melanocytes, when observed in situ have extensive dendritic branches with dendritic spines similar to neurons. When co-cultured with primary human keratinocytes, they have local Ca2+ transients within these spines and within the dendrites. These are elicited by secreted factors from adjacent keratinocytes. Thus other cell types with dendrites are capable of compartmentalized Ca2+ fluctuations in response to cell-cell communication. Furthermore, our observations within intact human skin suggest a more complex communication network between adjacent melanocytes and keratinocytes, and thus a more complex physiology to skin than previous appreciated.

This high percentage of melanocytes with Ca 2+ transients (64%±11%) was only observed when melanocytes were 125 co-cultured with keratinocytes. Fewer melanocytes had Ca 2+ transients when grown with unlabeled melanocytes 126 (16%±2%) or with HEK293T cells (13%±6%) ( Figure 4A). The presence or frequency of these calcium transients was not 127 affected by the growth media formulation (Figure 4 Supplement 1A, B). Furthermore, the number of melanocytes with 128 Ca 2+ transients did not increase when melanocytes were grown in keratinocyte-conditioned media nor when separated 129 from keratinocytes by a semi-porous membrane (Figure 4 Supplement 1C, D). Thus, close proximity to, and possibly 130 direct contact with, keratinocytes was required for the Ca 2+ transients. 131 To determine the source of Ca 2+ responsible for the dendritic transients, we quantified the number of Ca 2+ 132 transients before and after the removal of external CaCl 2 and/or addition of 1 µM thapsigargin, which releases calcium 133 from internal stores (Lytton et al., 1991). The percent of cells with Ca 2+ transients was reduced by either the removal of 134 external CaCl 2 or addition of thapsigargin (63±4% and 30±5% of pretreatment level, respectively) compared to the control 135 (88±5% of the pretreatment, Figure 4B), with the combination of removal of CaCl 2 and addition of thapsigargin having the 136 greatest effect (7±4% of the pretreatment level). This was also true for the number of Ca 2+ transients per cell ( Figure 4  137 Supplement 2). 138 139 ET-1, a keratinocyte derived paracrine factor, contributes to dendritic Ca 2+ transients 140 141 Endothelins, a family of 21 amino acid peptides, play a critical role in melanocyte development, maturation, and 142 homeostasis within the epidermis (Reid et al., 1996). Epidermal keratinocytes produce and secrete endothelin 1 (ET- The localized Ca 2+ transients in melanocytes elicited by ET-1 ( Figure 4C and Movie 5) resembled those seen in 152 co-cultures of melanocytes and keratinocytes without the addition of exogenous ET-1 (Figure 3 and Movie 1). To test if 153 the local Ca 2+ transients in melanocytes resulted from release of endothelin by keratinocytes, we treated co-cultures with 154 endothelin receptor antagonists. Co-cultures treated with the ET B antagonist had a significant reduction in the percent of 155 melanocytes with Ca 2+ transients but no change in Ca 2+ transients in co-cultures treated with the ET A antagonist ( Figure  156 4F). To confirm that the effects were due to inhibition of the ET B on melanocytes and not the consequence of inhibition of 157 ET B on keratinocytes, we used DsiRNA to knock down either ET A or ET B in melanocytes prior to co-culture with 158 keratinocytes. Consistent with the endothelin receptor antagonist data, knock down of ET B but not ET A , reduced the 159 number of co-cultured melanocytes with Ca 2+ transients ( Figure 4F).

161
Heterogeneous melanocyte response to acetylcholine. 162 163 Inhibition of ET B did not completely eliminate all Ca 2+ transients in co-cultured melanocytes. In addition to 164 producing and releasing endothelin, keratinocytes also possess the necessary cholinergic machinery to synthesize, 165 release and degrade acetylcholine (ACh), a classic neurotransmitter (Grando et al., 1993). Furthermore, melanocytes 166 express muscarinic receptors (mAChR) and respond to the mAChR specific agonist, muscarine (Buchli et al., 2001).  Supplement 6B). ACh increased the frequency of calcium transients, in those cells that already showed activity, but had 172 no effect of number of melanocytes with Ca 2+ transients (up to100µM, Figure 4 Supplement 5B). Atropine decreased the 173 number of co-cultured melanocytes with transients to 54±8% of the pretreatment value ( Figure 4I) parallel between melanocytes and cells of the nervous system which has profound implications for how melanocytes 186 receive and process information from their surroundings. It also suggests that there may be a more complex 187 communication network between melanocytes and keratinocytes. These results show that one melanocyte, through its 188 dendritic arbor, physically interacted with multiple keratinocytes and keratinocytes contacted multiple melanocytes. This 189 suggests that the melanocyte-centered "units" within the epidermis may not be independent. 190 Furthermore, we found that keratinocytes also have small cytoplasmic projections both in-situ and in culture which 191 made contact with melanocytes ( Figure 1G,H and Figure  1998). Here were show that in co-culture with keratinocytes, melanocyte dendritic Ca 2+ transients originated from distinct 211 regions within the melanocyte dendrite with subsequent events occurring within the same region as well as at specific 212 dendritic spines (Figure 3 ). This was also true of ET-1 and ACh induced Ca 2+ transients (Figure 4 Supplement 3 -6). 213 These data are indicative of functional domains along melanocyte dendrites in which receptors and linked signaling 214 cascades are restricted to, or active at, specific regions of the dendrite. However, further investigation is required to 215 determine exactly what factors contribute to the observed spatial distribution of localized Ca 2+ transients, including, but not 216 limited to, Ca 2+ buffering proteins, ER Ca 2+ availability, spatial distribution of IP3 receptors, as well as store operated 217 calcium and calcium induced calcium response machinery. 218 Through its dendritic arbor, one melanocyte physically interacts with multiple keratinocytes. The presence of 219 functional domains across the entire dendrite would allow melanocytes to locally modulate cell processes within a specific 220 region of the dendrite. One possible function for this type of spatial control could be local regulation of the production and 221 transfer of melanin. The observation that, on average, a higher percentage of Ca 2+ transients were observed in 222 melanocytes from derived from dark skin than from light skin ( Figure 3 Supplement 2) suggests they might regulate local 223 melanin production and/or melanosome transfer. This is consistent with previous work that showed an overall increased 224 cytoplasmic Ca 2+ enhances mono-cultured melanocytes' response to melanogenic stimuli (Carsberg et al., 1995).

226
Melanocyte heterogeneity 227 Interestingly, not all melanocytes responded to ACh with detectable increases in Ca 2+ transients, which indicates 228 that at the individual level there is heterogeneity in response to ACh ( and co-culture lead to Ca 2+ transients across the entire melanocytes ( Figure 6C and Figure S6A,E). increases, offers a possible mechanism by which melanocytes can make decisions regarding such outcomes.

647
Cell Culture 648 Melanocytes and keratinocytes were grown in melanocyte growth media (Medium 254 with Human Melanocyte Grow 649 Serum-2 (HMGS-2), Gibco) and keratinocyte growth media (Epilife with Human Keratinocyte Growth Serum (HKGS), 650 Gibco), respectively. Only cells 5 passages or less were used for experiments. 293T-HEK cells were grown in DMEM, 651 high glucose, pyruvate media (Gibco) supplemented with non-essential amino acids (Gibco) and 10% FBS at 652 37˚C, 5% CO 2 . 653 654 Optimized Melanocyte -Keratinocyte Co-culture System 655 Building upon previous studies (33-35), we optimized a melanocyte-keratinocyte co-culture to mimic physiological 656 conditions that also allowed us to generate the culture within a few days and use standard wide-field fluorescence 657 microscopy techniques.

658
A confluent multi-layer sheet of melanocytes and keratinocytes was achieved by seeding at a high density (0.5x 659 10 4 melanocytes and 5x10 4 keratinocytes per 78 mm 2 ) and allowing the cells to adhere and grow at 37˚C, 5% CO 2 for 17-660 24 hours in keratinocyte growth media (EpiLife media with HKGS). This media formulation is used for mono-culture of 661 primary human keratinocytes without the presence of a fibroblast feeder layer and contains 0.06mM CaCl 2 . This lower 662 than physiological CaCl 2 is used to maintain keratinocytes in an undifferentiated state which allows them to proliferate 32,33 . 663 Melanocytes will grow and survive in keratinocyte media, in the presence of keratinocytes. However, both cell-cell 664 adhesion and keratinocyte differentiation is Ca 2+ dependent 34 . Co-cultures kept at low CaCl 2 remained as a monolayer, 665 did not form adherens junctions and exhibited an altered cytoskeletal morphology that differed from healthy human skin 666 (Figure 2 Supplement 1A). To achieve a multi-layer culture with proper cell-cell adhesion, cell morphology and both 667 undifferentiated and differentiating keratinocytes, we increased the CaCl 2 to 1.06 mM in the growth media after the initial 668 monolayer was formed. This restored adherens junctions and supported a more physiological actin cytoskeleton structure